Innovation System Frontiers: Cluster Networks and Global Value (Advances in Spatial Science)

Advances in Spatial Science Editorial Board Manfred M. Fischer Geoffrey J.D. Hewings Peter Nijkamp Folke Snickars (Coordinating Editor)

Titles in the Series M. M. Fischer, J. Revilla Diez and F. Snickars Metropolitan Innovation Systems VIII, 270 pages. 2001. ISBN 978-3-540-41967-9

R.J.G.M. Florax and D. A. Plane (Eds.) Fifty Years of Regional Science VIII, 400 pages. 2004. ISBN 978-3-540-22361-0

L. Lundqvist and L.-G. Mattsson (Eds.) National Transport Models VIII, 202 pages. 2002. ISBN 978-3-540-42426-0

D. Felsenstein and B.A. Portnov (Eds.) Regional Disparities in Small Countries VI, 333 pages. 2005. ISBN 978-3-540-24303-8

J. R. Cuadrado-Roura and M. Parellada (Eds.) Regional Convergence in the European Union VIII, 368 pages. 2002. ISBN 978-3-540-43242-5

A. Reggiani and L.A. Schintler (Eds.) Methods and Models in Transport and Telecommunications XIII, 364 pages. 2005. ISBN 978-3-540-25859-9

G. J. D. Hewings, M. Sonis and D. Boyce (Eds.) Trade, Networks and Hierarchies XI, 467 pages. 2002. ISBN 978-3-540-43087-2 G. Atalik and M. M. Fischer (Eds.) Regional Development Reconsidered X, 220 pages. 2002. ISBN 978-3-540-43610-2 Z. J. Acs, H. L. F. de Groot and P. Nijkamp (Eds.) The Emergence of the Knowledge Economy VII, 388 pages. 2002. ISBN 978-3-540-43722-2 R. J. Stimson, R. R. Stough and B. H. Roberts Regional Economic Development X, 397 pages. 2002. ISBN 978-3-540-43731-4 S. Geertman and J. Stillwell (Eds.) Planning Support Systems in Practice XII, 578 pages. 2003. ISBN 978-3-540-43719-2 B. Fingleton (Ed.) European Regional Growth VIII, 435 pages. 2003. ISBN 978-3-540-00366-3 T. Puu Mathematical Location and Land Use Theory, 2nd Edition X, 362 pages. 2003. ISBN 978-3-540-00931-3 J. Bröcker, D. Dohse and R. Soltwedel (Eds.) Innovation Clusters and Interregional Competition VIII, 409 pages. 2003. ISBN 978-3-540-00999-3 D. A. Griffith Spatial Autocorrelation and Spatial Filtering XIV, 247 pages. 2003. ISBN 978-3-540-00932-0 J. R. Roy Spatial Interaction Modelling X, 239 pages. 2004. ISBN 978-3-540-20528-9 M. Beuthe, V. Himanen, A. Reggiani and L. Zamparini (Eds.) Transport Developments and Innovations in an EvolvingWorld XIV, 346 pages. 2004. ISBN 978-3-540-00961-0 Y. Okuyama and S. E. Chang (Eds.) Modeling Spatial and Economic Impacts of Disasters X, 323 pages. 2004. ISBN 978-3-540-21449-6 L. Anselin, R.J.G.M. Florax and S. J. Rey Advances in Spatial Econometrics XXII, 513 pages. 2004. ISBN 978-3-540-43729-1

H.W. Richardson and C.-H.C. Bae (Eds.) Globalization and Urban Development X, 321 pages. 2005. ISBN 978-3-540-22362-7 G. Arbia Spatial Econometrics XVII, 207 pages. 2006. ISBN 978-3-540-32304-4 B. Johansson, C. Karlsson, R. Stough (Eds.) The Emerging Digital Economy X, 352 pages. 2006. ISBN 978-3-540-34487-2 H.Westlund Social Capital in the Knowledge Economy X, 212 pages. 2006. ISBN 978-3-540-35364-5 A.E. Anderssson, L. Pettersson, U. Strömquist (Eds.) European Metropolitan Housing Markets VI, 363 pages. 2007. ISBN 978-3-540-69891-3 A.T. Murray, T.H. Grubesic (Eds.) Critical Infrastructure VIII, 311 pages. 2007. ISBN 978-3-540-68055-0 R. Cooper, K. Donaghy, G. Hewings (Eds.) Globalization and Regional Economic Modeling XIII, 475 pages, 2007. ISBN 978-3-540-72443-8 R. Capello, R. Camagni, B. Chizzolini, U. Fratesi Modelling Regional Scenarios for the Enlarged Europe: European Competiveness and Global Strategies XX, 337 pages, 2008. ISBN 978-3-540-74736-9 C. Jensen-Butler, B. Sloth, M. Larsen, B. Madsen and O.A. Nielsen Road Pricing, the Economy and the Environment XVI, 425 pages, 2008. ISBN 978-3-540-77149-4 U. Fratesi and L. Senn (Eds.) Growth and Innovation of Competitive Regions X, 370 pages, 2009. ISBN 978-3-540-70923-7

Brian Wixted

Innovation System Frontiers Cluster Networks and Global Value

Dr. Brian Wixted The Centre for Policy Research on Science and Technology Simon Fraser University 515 West Hastings Street Vancouver, B.C. Canada V5A 1S6 [email protected]

ISBN 978-3-540-92785-3

e-ISBN 978-3-540-92786-0

DOI: 10.1007/978-3-540-92786-0 Library of Congress Control Number: 2008944000 © Springer-Verlag Berlin Heidelberg 2009 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Cover design: WMW Design, Heidelberg Printed on acid-free paper springer.com

To Bonnie For her love and encouragement.

Acknowledgments

This book is largely based on my doctoral research, carried out at the University of Western Sydney’s AEGIS Research Centre, although it has been significantly extended here with time series data and additional contextual analysis included in the latter chapters. I want to thank my Doctorate supervisors Professors Jane Marceau, Tim Turpin and Russel Cooper for their advice, encouragement and editorial assistance. Professor Cooper in particular gave important assistance with the development of modelling software and many hours of tutoring in matrix algebra. Colleagues, Professor Maureen McKelvey, Dr Mike Hales were both particularly influential and helpful in guiding my early research path. The words of wisdom from the doctoral examination panel of Professors Stan Metcalf, Keith Smith and Geoffrey Hewings contributed significantly to the current volume. More recently, Adam Holbrook has given important guidance. Earlier, during my employment with the Australian public service, a number of friends and colleagues encouraged my activities and work in the field of science and technology policy and studies, so here I would like to acknowledge Dr Kevin Bryant, Ian Tranter, Stuart Smith, Chris Birch, Julian O’Dea and John Madden. I hope this work is in itself some reward. To my friends, Greg Restall, Christine Parker, Bruce and Faye Spencer, Gail Stevenson, Jennifer Collier, Shane and Nerida Gill, Ian and Elizabeth Campbell, Rob and Heather Sturgiss and Judith Carpenter amongst many, go my warm thanks for their friendship and encouragement. Thanks also to Angela Paloubis for her editorial assistance, Katharina WetzelVandai and other staff at Springer and to Geoffery Hewings as editor of Advances in Spatial Sciences series for his encouragement to publish this research. The usual disclaimers apply. Canada February 2009

Brian Wixted

vii

Preface

The concept of innovation systems has come to dominate the discourse on technological innovation over the last two decades of research. This book is devoted to pushing at the frontiers of this area of knowledge both metaphorically by contributing new theoretical and empirical insights, and also through its content focus on the geographic frontiers that have been fixed within the literature to delineate one ‘system’ from another. Thus, my interest here is in frontier as boundary, border, and jurisdiction: politically, industrially and geographically. The overarching theme behind the research reported here is the basic concept that at least one sub-field of the study of innovation systems should focus on a multi-spatial framework which facilitates analysis of how places are connected to one another. A strong finding on innovation is that it is a systemic property of groups of businesses, not lone inventors. However, it is common to restrict systems to nation states, sub-national regional political jurisdictions or intra-country industrial clusters. These clusters of activities are not just conceived as national, but are often analysed as enclaves with little reference to their relationships with the rest of the world. This book attempts to push at these frontiers and by using large inter-country input-output datasets to show the extent of international networks of production. Results presented here indicate that some technologically complex products involve a high degree of component transfer across borders. When these production systems are further examined (auto in chapter seven, civil aerospace in chapter eight and ICT in chapter nine) it can be shown that international cluster to cluster relations remain fairly stable across time with new nodes (clusters) often being included into existing networks rather than displacing existing ones. However, the degree of dependence on imports is growing. The findings on the patterns and characteristics of these external networks (or cluster complexes as they have been called here) bear a striking similarity to the findings of within-cluster dynamics.

ix

Contents

1

2

3

Introduction ................................................................................................

1

1.1 1.2

The Economics of Regionalisation vs. Globalisation ......................... Bridging the Gap: Places and Extra-territorial Spaces ........................ 1.2.1 The Advantages of Place: Innovation Systems ....................... 1.3 The Fragmentation of Production – Despatialisation? ........................ 1.3.1 Constructing a Cluster Complexes Framework ....................... 1.4 Specifying the Research Problem........................................................ 1.4.1 The Key Questions .................................................................. 1.4.2 The Structure of the Book ....................................................... 1.4.3 Boundaries and Contributions of this Research ...................... 1.5 Overview .............................................................................................

1 2 2 4 5 6 7 8 10 11

Systemic Innovation and Nation States....................................................

13

2.1 2.2

Knowledge, Technology, Innovation and Nation States...................... Technology, Competitiveness and Systems ......................................... 2.2.1 Economic Growth .................................................................... 2.2.2 Innovation Aids All Sectors from Agriculture to Services...... 2.2.3 Innovation Embedded in Nations ............................................ 2.2.4 Systems Theory ....................................................................... 2.3 The Historical and Continuing Importance of Nations ....................... 2.3.1 Learning Nations ..................................................................... 2.3.2 Production System Specialisation and Trajectories................. 2.4 Nations in a Changing World: Weaknesses of NIS ............................. 2.4.1 NIS and National Geography .................................................. 2.4.2 NIS Politically Defined............................................................ 2.4.3 NIS, Borders and Economic Space .........................................

13 14 14 16 18 19 23 23 27 28 29 30 32

Innovative Regions, Clusters and Milieux ...............................................

33

3.1 3.2

33 35 36 37

Innovation Territories .......................................................................... Locating Clusters................................................................................. 3.2.1 Clusters, Regions and Sectors ................................................. 3.2.2 Defining and Finding the Places of Interest ............................

xi

xii

4

5

Contents

3.3 The Geography of Agglomeration ...................................................... 3.3.1 Knowledge and Innovation Geography ................................... 3.3.2 Supply Architecture ................................................................. 3.4 Cluster Based Theories of the Drivers of Proximity ........................... 3.4.1 Traded Interdependencies: Supply Architectures .................... 3.4.2 Users and Producers: Linking Production and Innovation ...... 3.4.3 Untraded Interdependencies: Knowledge Flows ..................... 3.5 Cluster Boundaries and Extra-territorial Linkages.............................. 3.5.1 Contiguous Regions................................................................. 3.5.2 Clusters Beyond Proximity...................................................... 3.6 Multi-spatial Production and Innovation Spaces .................................

39 40 42 44 45 48 50 52 53 54 57

Beyond Borders: Trade and Networks.....................................................

59

4.1 4.2

Traded Interdependencies Beyond Borders......................................... Trade, Borders and the Sourcing of Products...................................... 4.2.1 Trade Specialisation Patterns................................................... 4.2.2 Bilateral Trade Relations ......................................................... 4.2.3 International Value Chains and Spatial Specialisations........... 4.3 Proximity and Multi-Spatial Innovation Systems ............................... 4.3.1 Spatial Innovation Systems: NIS and Clusters ........................ 4.3.2 The Multi-Spatial Gap: Cluster Context and Trade Linkages ................................................................. 4.3.3 Linked Clusters: Specialised Nodes, Globalised Products ...... 4.4 Clusters, GPNs or Multi-Spatial Cluster Networks? ........................... 4.4.1 Australia’s Aerospace Micro-Cluster ......................................

59 60 61 63 71 78 78

Measuring Inter-Cluster Interdependencies ...........................................

89

5.1

Identifying Multi-Spatial Innovation Systems .................................... 5.1.1 The Appeal of I-O Modelling .................................................. 5.1.2 Criticisms of Using I-O in Innovation Studies ........................ 5.1.3 Benefits of I-O Modelling ....................................................... 5.2 Multi-Country Input-Output Data ....................................................... 5.3 Inter-Regional I-O Modelling.............................................................. 5.3.1 Constructing the I-O Tables..................................................... 5.3.2 Limitations of Trade Data........................................................ 5.3.3 Inter-Country Modelling Analyses .......................................... 5.3.4 Net Multipliers......................................................................... 5.4 Data Analysis: Chaps. 6–9 .................................................................. 6

80 81 83 83

89 90 91 92 96 97 97 101 102 103 105

Clustering Internationalisation................................................................. 107 6.1 6.2

Dependence on Imported Components ............................................... 107 Complexity, Clustering and Imports ................................................... 113 6.2.1 Knowledge Economy and Clustering ...................................... 113

Contents

6.2.2 R&D Intensity ......................................................................... 6.2.3 Pavitt’s Industry and Innovation Taxonomy ............................ 6.2.3 Complex Technologies ............................................................ 6.3 Imports, Science and Scale.................................................................. 6.3.1 Import Intensiveness ................................................................ 6.3.2 Implications for Theory, Research and Policy......................... 7

The Significance of Automotive Production Systems ......................... Background: A Traditional Perspective of Exports ............................. The Evolution of the Inter-cluster Networking ................................... 7.3.1 Auto Activities 1970................................................................ 7.3.2 Auto Activities 1990................................................................ 7.3.3 EU 15 1995 .............................................................................. 7.3.4 Auto Activities 2000................................................................ 7.4 Techno-Organisational Context of Auto Production ........................... 7.4.1 Clusters, Complexes and System Hierarchies ......................... 7.4.2 Geography and Organisation ................................................... 7.4.3 Reconfiguration ....................................................................... 7.5 Implications and Conclusions .............................................................

125 126 129 129 130 132 135 139 139 141 142 145

Cluster Complexes: Civil Aerospace ........................................................ 147 8.1 Aerospace: The Techno-Organisational Context................................. 8.2 Cluster Networking ............................................................................. 8.2.1 Aerospace 1970 ....................................................................... 8.2.2 Aerospace 1990 ....................................................................... 8.2.3 Aerospace 1995–2000 ............................................................. 8.2.4 Aerospace Import Changes...................................................... 8.3 The Beginnings of the Future of Aerospace Geography ..................... 8.4 Implications and Conclusions ............................................................. 8.4.1 Innovation Theory and Analysis.............................................. 8.4.2 Policy Matters ..........................................................................

9

114 117 120 123 123 124

Cluster Complexes: Auto Production....................................................... 125 7.1 7.2 7.3

8

xiii

147 149 149 151 151 155 156 161 162 162

Cluster Complexes: Electronics and ICT ................................................ 165 9.1 9.2

Introduction: Structures of Interdependencies .................................... The Third Technological Revolution: ICT .......................................... 9.2.1 Economic Growth and ICT...................................................... 9.2.2 ICT Clustering ......................................................................... 9.2.3 ICT Exports After 1990 ........................................................... 9.3 ICT Cluster Networks ......................................................................... 9.3.1 ICT 1970.................................................................................. 9.3.2 ICT 1990.................................................................................. 9.3.3 ICT 1990–2000........................................................................

165 166 166 167 169 171 171 173 173

xiv

Contents

9.4

9.5 10

Contemporary and Emerging Nodes of Global ICT ....................... 9.4.1 The Techno-Organisational Context in ICT ...................... 9.4.2 Contemporary East Asian Nodes....................................... 9.4.3 Emergent Nodes in the Global Architecture of Production ..................................................................... Conclusions and Implications .........................................................

181 182 183 187 189

Conclusions on the Architecture of Economies ..................................... 191 10.1

Atolls of Innovation or Something Else? ........................................ 10.1.1 Local Agglomeration and Specialisation........................... 10.1.2 Bridging Local and Global: The Role of Interdependencies .......................................................... 10.1.3 The Scale of Connectedness – Do These Links Matter?..................................................................... 10.1.4 Innovation Atolls? ............................................................. 10.2 Systems of Systems: Sectoral Footprints ........................................ 10.2.1 The Spatial Structure of Clustering and Networking ........ 10.2.2 Clustering, Fragmentation and Integration – Economic Fractals ............................................................. 10.2.3 Cluster Complexes: Nodes, Flows and Hierarchies .......... 10.2.4 Spatial Organisation: Assembly, Integration and Technologies ............................................................... 10.3 Conclusions and Implications ......................................................... 10.3.1 Clusters Don’t Innovate in Isolation .................................. 10.3.2 Implications for Theory ..................................................... 10.3.3 Implications for Policy ...................................................... 10.3.4 Future Research .................................................................

191 193 194 195 196 197 197 197 198 199 200 201 202 202 204

References ......................................................................................................... 205 Index .................................................................................................................. 225

Chapter 1

Introduction

‘Although there is a large literature on the internationalization of economic activity (including R&D) at the corporate level, there are relatively few studies of the degree of internationalization of innovation systems’ (Carlsson 2006, p. 64).

1.1

The Economics of Regionalisation vs. Globalisation

Over the past 20 years or so, the landscape of economic development around the world has been changing remarkably quickly. The continued economic advance of Asian countries like South Korea, Taiwan, and Singapore; the emerging strength of China; the new software development hotspots in India; and the movement of factories to countries like Malaysia and Mexico point to significant changes in the organisation of industrial production. Such changes, on the ground, highlight a growing divide in the literature on the factors for successful economic development and industrial location. Some authors emphasise the significance and role of the nation and national institutions for the development of technological capability and innovation across an economy. Others dismiss the nation state as increasingly irrelevant with waning powers to influence development trajectories in a globalised economy. Understanding even a few more of the critical elements of the processes leading to the visible structural changes would be valuable for devising policies in advanced, emerging and less developed economies alike. The academic literature on technological innovation has not only argued for the importance of the generation of knowledge to industrial competitiveness and economic growth, but also for the importance of national characteristics for the creation of that knowledge. Freeman (2002) and Lundvall et al. (2002) continue to argue for the focus of analytical effort to remain on national innovation systems as the set of social arrangements in which learning and the development of technological capability principally occurs. This field of research emphasises the systemic nature of knowledge, technology and product and process innovations in successful industry evolution, rather than traditional industrial structure, micro and macroeconomics analyses. As components of nation states, the so-called ‘sub-national’ systems of innovation, which include clusters and regions, is now an established B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_1, © Springer-Verlag Berlin Heidelberg 2009

1

2

1

Introduction

area of interest, which whilst accepting the basic principles of the national systems approach nevertheless, takes a different spatial frame, even if that is within the borders of nations. As Carlsson states: ‘in view of the fact that most studies of innovation systems focus on national innovation systems, it is not surprising that little direct evidence is found that innovation systems are becoming global. The main focus in this literature is on institutions at the national level. But national institutions may influence innovation systems at regional, sectoral, or technological levels differently. However, at these lower levels there has been little work done with a view toward internationalization of systems (as distinct from corporate innovative activity). Also, not all institutions are national. For large firms, national institutions may be most important, while for small and new firms, sub-national institutions may also be important’ (2006, p. 65).

Alongside this perspective on the role of nations is an extensive discussion in other literatures on the fragmentation of production processes across borders, the growing levels of international trade and flighty foreign direct investment (FDI) capital that appears to be moving easily around the world in search of opportunities for generating high returns. This focus on the international division of labour based on costs and the shifting geography of manufacturing activities has led many researchers to conclusions which are opposite to those of the researchers studying innovation. The patterns of increasing trade can be interpreted as the weakening of the nation state and for some even as its demise (Ohmae 1995). Thus, the term ‘globalisation’ (e.g. amongst many others; Hatzichronoglou 1999, Yeats 1998, and Ng and Yeats 1999) has entered the popular vocabulary, signifying not simply trade patterns, but also the increasing levels of foreign ownership of assets and the expansion of the role of international agreements that constrain a variety of government policy actions. The division between analysis that emphasises the importance of location and the flows of trade and the relocation of production needs to be bridged. As Sturgeon comments: ‘We need to better understand the various roles that local agglomerations play within spatially extensive value chains and begin to map the activities that tend to concentrate in particular places even as the geographic ‘footprint’ of linked economic activity expands. (2003, p. 200).

Thus, in the existing research, these two perspectives of clustering and fragmentation (as it has been called) have been presented as diametrically opposed processes, as though they could not be occurring simultaneously. The aim of this book is to explain why these processes, rather than being opposed to one another, appear to be self-reinforcing.

1.2 1.2.1

Bridging the Gap: Places and Extra-territorial Spaces The Advantages of Place: Innovation Systems

Since the beginning of the 1990s, it has been increasingly clear that technological innovation has an important role in economic development and growth. The think tank of world’s wealthiest countries, the Organisation for Economic

1.2

Bridging the Gap: Places and Extra-territorial Spaces

3

Co-operation and Development (OECD), has been at the forefront of promoting this view, publishing a number of influential studies (see for example OECD 1992, 1996c and 2001b). Within the study of technological innovation, sometimes referred to as neo-Schumpeterian, for the contribution Josef Schumpeter made to positioning innovation at the centre of economic processes, or alternatively as the economics of innovation and technological change (EITC) there are many important research themes currently being investigated. From analyses of individual technologies, corporate activities (knowledge management and technology creation), and government policy interventions and through to the analyses of the innovativeness of entire countries (e.g. see Nelson 1993), a vast array of topics are being discussed and debated in the international literature. Increasingly, a large proportion of the literature is influenced by geography-based analysis of the systemic characteristics of innovation (national and regional innovation systems). The insight that innovation is systemic and thus, that places are important sites of economic advantage has been a critical change in thinking. Businesses, far from being purely independent actors, rest on a series of foundations of knowledge generation institutions (research organisations and universities), cultural influences on learning, knowledge dissemination organisations (schools and universities); government policies and the network of other firms that exist in particular places (see Chaps 2 and 3). These system variables influence the trajectories of ‘national innovation systems’, which Patel and Pavitt (1994) have shown to be quite stable, relative to other countries, over the long term. There are now a number of different ‘system’ levels of analysis, including: • National innovation systems (nation states); • Regional innovation systems (states, provinces – substantial but sub-national politico-administrative geographies); • Sectoral innovation systems (emphasises corporate and public organisations contributing to particular production systems which in practice tend to (but need not necessarily) be limited by nation state borders; and • Clusters (these extend over various spatial scales but emphases knowledge and production systems). Of the sub-national themes, the concept of clustering has become the more popular term and spans both the economic geography and EITC fields of studies. The OECD has played a role in bringing together cluster case studies (OECD 1999d and 2001c) and there have been at least three major research projects on clustering which have investigated a significant number of case studies. The Innovation Systems Research Network* has concluded a study of 26 ‘clusters’ in Canada. These ranged in size from small local clusters to large clusters of national importance (auto), from the high technology (biotechnology) through to lower technologies (food), and all being examined with a common methodology (in particular Wolfe and Gertler 2004). IT clusters in Taiwan, USA, the UK, Israel and Scandinavia were the focus of a group of researchers that included Saxenian (e.g. see Bresnahan et al. 2001). Finally, the Institute for Strategy and Competitiveness’ *See Holbrook and Wolfe 2000, Holbrook and Wolfe 2002, Wolfe 2003 and Wolfe and Lucas 2004.

4

1

Introduction

Cluster Mapping Project at Harvard University is analysing the location quotients and growth performance of clusters in the USA (see Porter 2003). This and other research has already played a valuable role in advancing our knowledge of the characteristics of local economic agglomerations.

1.3

The Fragmentation of Production – Despatialisation?

The fragmentation approach pioneered by Jones and Kierzkowski (1990 and 2001) and Arndt (1998) is an area of research that focuses upon the process by which production is fragmenting with an increasing number of processing stages, in particular, when it is organised across more than one location. The two key drivers of this fragmentation process, as these and other like-minded authors perceive it, is driven by reductions in transportation and communications costs that have opened up access to new (cheaper) locations for production. As Kierzkowski states: ‘What is new, is the increasing importance of the off-shore element in outsourcing and the growing role of low-wage countries in that development. What has made this growth possible has been the recent revolution in communications and related technologies and a sharp reduction in coordination costs that came with it. With the death of distance, to borrow the title of a recent book in this area, the scope for modularizing and reorganizing production processes has increased considerably. No wonder that the countries of East Asia have been exploiting, in a good sense of the word, new opportunities’ (emphasis added 2001a, p. 7).

To date, a large number of the articles in this area have adopted a highly theoretical modelling based approach and exhibit a particular interest in the welfare effects of changes both in developed and developing economics. ‘Economists have investigated this phenomenon with a focus on welfare and factor-price effects, mainly using Heckscher-Ohlin-type trade models. Existing studies emphasize a positive welfare effect of international fragmentation, but reveal ambiguous effects on factor prices’ (Kohler 2001, p. 31).

Another strand of this research is based on very specific case studies of individual situations where the production process has geographically fragmented. For example, there is analysis of bicycles (Chu 2001), hard disk drives (Kimura 2001) and others. Currently, there is only a nascent systems level empirical development (see Feenstra 1998, Hummels et al. 2001 and Chen et al. 2005) of the fragmentation thesis. As such, there is little evidence presented on the degrees to which different industries are dependent on imports, nor differentiation between re-localisation and fragmentation. Researchers thus seem to have overlooked the evidence that it is both particular resource processing and high technology activities that have been fragmenting the most in OECD economies (see Table 6.1). Further, and notwithstanding, East Asia’s ability to capture a significant share of activity in electronics, traditional advanced economies are still, in many instances, deepening their interdependence on each other.

1.3

The Fragmentation of Production – Despatialisation?

1.3.1

5

Constructing a Cluster Complexes Framework

In researching how, simultaneously, innovation is systemic and place oriented, while production processes are fragmenting in the modern economy, both perspectives would be of equal value as a starting point. The systems of innovation literature has largely overlooked the fragmentation of industries, yet the fragmentation literature has largely overlooked the characteristics of products (technologies) as being a key variable in the fragmentation of production. Therefore, it is critical to understand the primary characteristics that are driving and shaping these industries and their development. Based on this need, the decision here has been to focus the book primarily on the systemic features of innovation and role of place in shaping innovation. However, other relevant research literatures are discussed and analysed, see in particular Chap. 4. Carlsson (2006) suggests that neo-Schumpeterians do not have a developed language for describing the global scale of economic structures or the international system of innovation systems. So, while economic geographers are interested in addressing the issue of global production, even if they think it has not been dealt with adequately (Dicken 2004) or they are setting out a research agenda for the ‘spatial logic of economic development’ (Tellier 1997), innovation is still contextualised within political jurisdictions. This is despite the disparity in national systems, ranging for advanced economies, from the United States of America (USA) with 291 million people (the largest OECD country) and Iceland with just 289,000 people (the smallest) (data from OECD 2004, pp. 6–7). Analysts of national and sub-national innovation systems have often adopted a research methodology based on endogenous capability (e.g. see Lundvall 1992, Nelson 1993 and see Chaps. 2 and 3), as opposed to developing models of crossspatial interdependency. As Breschi and Malerba comment, ‘a key feature of successful high-technology clusters is related to the high level of embeddedness of local firms in a very thick network of knowledge sharing’ (2001, p. 819) with firms able to tap into a ‘body of localized knowledge and capabilities’ (p. 820). This paradigm seems to prevail even when a particular set of activities is explicitly conceived of as a ‘node’ within a global value chain. The ‘thickness’ of relations with local institutions was used by Henry and Pinch (2001) to determine the ‘embeddedness’ of the British auto ‘valley’ with little, if any, determination of its nodal dimensions within global knowledge and industrial networks. While much of the literature on clustering and economic agglomeration has focused upon particular places as if they were not connected to the rest of the world, recent evidence emphasises that the national and international linkages1 maintained by cluster-based firms are vital sources of knowledge. Bunnell and Coe (2001), Oinas and Malecki (2002), Wolfe and Gertler (2004) as well as Simmie (2004) all take an interest in the take in the linkages across borders. Simmie, for example,

1

Giuliani et al. (2005) focus on the importance of exports for clusters.

6

1

Introduction

notes that leading innovators in the United Kingdom rely more on national and international linkages than on local and regional linkages (2004, p. 1105). Meanwhile, Wolfe and Gertler (2004) take issue with the Porterian concept that clusters require supply networks to be in close proximity. These articles are, however, the exception and as such are notable because they highlight the dearth of research on the connectedness of clusters (see Chap. 3 below). There are a few obvious examples of situations where clusters of all sizes are strongly linked in economic space but are either geographically distant or separated only by a national or politico-administrative border. Some easily identifiable examples include: • • • •

Detroit (USA) – Windsor (Canada) [auto clusters]; New York – London – Tokyo [world financial centres]; Silicon Valley – Bangalore [ICT clusters]; and Melbourne (Australia) – Adelaide (Australia), [sub-national auto clusters].

It has been widely suggested (see Chap. 4 below), that the assembly of final products is increasingly integrating components made or designed in many parts of the world. Everything from low cost products (such as textiles, clothing and footwear), to processing centred production activities (steel and aluminium) and complex products (electronics, vehicles and aerospace) utilise a high level of imported components and industrial ingredients (see Chap. 6). Our poor awareness of the knowledge and production geography of individual products is an increasingly critical gap, but, at least, this is covered at the corporate level by the global production networks analyses. The wider research agenda on innovation systems at the spatial level, therefore, needs to move beyond location and ask whether particular clusters are in specialised niches or are globally powerful. Are some clusters relatively independent or so dependent on products purchased extra-regionally that, in fact, it is a broader economic space spanning political jurisdictions that should be considered as the functional cluster? The advantage of conceiving of ‘functional clusters’, as geographies being linked within a production value chain, is that it neither diminishes the economic importance of locations nor ignores the wider changes in economic geography. It is not that local systems are unimportant, only that they need to be identified within the global architecture of production. As one example, as there are now only two major assemblers of commercial aircraft (Boeing and Airbus), aerospace clusters across the world need to be highly inter-linked with each other and the major assemblers. Mapping these linkages will reveal the extra-territorial cluster interdependencies or here called cluster complexes.

1.4

Specifying the Research Problem

The entire intellectual territory of what might constitute a research agenda on innovation systems beyond borders is far too broad to be considered here. What needs to be established, in the first instance, is whether the phenomena of connections

1.4

Specifying the Research Problem

7

between specific sub-national systems matters a great deal or is somewhat peripheral – i.e. what is the significance of cluster linkages? The idea of significance can be separated into a number of different components. Because the focus is on the connections between places, the ones of interest here are the role, the scale and the spatial structure of linkages and interdependencies. Importantly, it is necessary to determine whether the patterns of linkages and local strengths matter for the study of innovation. Whether the role, scale and structure of linkages matters will be established in this book, at least in a preliminary manner, by determining whether sub-national systems of innovation are linked in networks or develop such a wide spread of connections that there no discernable pattern of inter-cluster interaction. To conduct this type of research it is necessary to develop a multi-spatial perspective. This term is used throughout the present book to denote both an analytical frame of reference and the methodological tools necessary for the analysis of cluster-to-cluster relations in economic space and across geographic space. Multispatial analysis potentially includes any study that could identify linkages between different places, whether those connections are in the use of products, knowledge, patents or the flow of human capital in migration patterns.

1.4.1

The Key Questions

To achieve the goals of researching the role, scale and spatial structure of interactions, more specific research questions were developed. These were: 1. Does the ‘innovation systems’ agenda give primacy to political geography (nation states and regions within nations) over other possible frameworks including economic space, which may lead to more research on extra-territorial links and if so, why; 2. What is it about linkages between economic actors (relationships particularly between users and producers) within innovation theory that makes them so important for the development and diffusion of new products and services; 3. What is known about product and knowledge linkages that extend beyond the borders of particular innovation systems (especially industrial clusters); 4. In multi-spatial research of innovation systems, is interregional input-output modelling useful and can it reveal linkages between systems that are statistically important; 5. Does the structure of international input-output relations make sense when other evidence on knowledge flows, technological specialisations and national and sub-national systems of innovation are considered; and 6. What new insights into systems of innovation and the interdependencies between them emerge from understanding the spatial structure of linkages? These questions are covered within the following chapter structure.

8

1.4.2

1

Introduction

The Structure of the Book

In introducing a research agenda focussed on the linkages between clusters (henceforth labelled cluster complexes), not just as a theoretical proposition but also with an attempt at developing a substantial volume of data, it was thought necessary to examine, in detail, how the relevant literature addresses border issues as well as the issue of interdependencies inside them and across them. For this reason, the present work is built upon extensive analysis of three broad themes of research: National innovation systems (Chap. 2), sub-national innovation systems (Chap. 3) and international trade (Chap. 4) literatures are analysed with a focus on the role of knowledge, why location matters and the nature of industrial interdependencies. To assist the reader navigate the literature covered in the present thesis, the following diagram has been developed to help identify the topics of interest. It is repeated in a modified form at the beginning of Chaps. 2, 3 and 4. The components of Fig. 1.1 how they are incorporated into the book, and an outline of the contents of the chapters follows: Chapter 2: Systemic innovation and nation states. It has already been noted that business innovation is now understood to be an outcome of system characteristics, which have traditionally been understood as developing within nation states. Chapter 2 explores the way innovation contributes to growth and competitiveness. It also examines the strengths and weaknesses of a system perspective founded on the geo-political entities of nation states (National Innovation Systems – NIS). Chapter 3: Innovative regions, clusters & milleux. Although the spatial reference is different, the analytical themes of the sub-national systems of innovation research are similar to the concerns of NIS research. At the meso level (not companies

Fig. 1.1 A literature map of relevant themes for this research

1.4

Specifying the Research Problem

9

and not nations) the studies are more focussed on specific geographic locations which often specialise in particular production activities. The research literature emphasises the importance of knowledge flows between organisations within a limited spatial area, with only limited evidence on the nature and structure of extra-regional (intra-national or international) linkages. The organisation of traded and untraded interdependencies and user-producer relationships are of particular interest. Chapter 4: Beyond borders: trade and networks. In the first section of the chapter, research findings from the fields of EITC, neo-classical economics, input-output economics, global production networks, production fragmentation and world city networks are presented, as they relate to the evolution of the spatial structure of trade. Chapter 4 also analyses the social network explanations of trade, as the findings are significant for understanding the flow of innovation and knowledge between localities. The second section of the chapter assesses the evidence from the first part of the book, presented across Chaps. 2, 3 and 4, to draw conclusions on the likely role, scale and spatial structure of extra-regional linkages. The section also highlights the multi-spatial research agenda by examining the case of the emerging aerospace cluster in Australia. Chapter 5: Measuring inter-cluster interdependencies. Chapter 5 has two major sections. The first assesses the benefits and weaknesses of using an input-output methodology to map the scale and structure of relationships and includes a consideration of the compatibility of this approach with the aims of innovation systems research. The second section describes the construction of various inter-country input-output models used in the research to analyse the flows of intermediate goods between industries and countries. The models include one based on data for 15 European Union (EU) countries for the year 1995 (Eurostat 2000) and a series based on OECD data for 1970, 1990, the mid 1990s and for 2000 (OECD 1995, 2002e and 2006). Chapter 6: Clustering internationalisation. Input-output modelling has a number of important advantages. One of them is that it enables a calculation of the value of purchases of industrial ingredients from different industries (in different countries) as a share of output. The first section of Chapter six presents data from the various models on those industries that require the highest levels of imported ingredients. The second section of Chap. 6 focuses on high import industries and, in particular, those that have a higher R&D intensity. The particular interest is whether measures of technological complexity can help account for the high import higher R&D clusters. Chapter 7: Cluster complexes in the automotive production system. One of the central themes of this book is whether clusters form spatially structured external relationships or whether linkages are dispersed across a wide variety of other systems. Chapter seven presents an analysis of the interdependencies between national transport equipment clusters (EU model) and motor vehicles clusters (OECD model). The chapter marries analysis of the input-output data with some of the available evidence on the technological and trade strengths of these industries in relevant countries.

10

1

Introduction

Chapter 8: Cluster complexes in the aerospace production system. The structure of chapter eight is similar to that of chapter seven, but focuses on the spatial organisation of the aerospace production system. The chapter again utilises both input-output modelling results and some of the available evidence on the technological and trade strengths of related industries in relevant countries. Chapter 9: Cluster complexes in the electronics production systems. Chapter nine has a similar focus and goal to that of chapters seven and eight and, therefore, has a similar structure. On this occasion, the interest is in cluster-to-cluster interdependencies for the office machines and computing and electronics within the EU and OECD model countries. As Asian countries have a strong position in the world economy in these sectors but the models used here cannot integrate them particularly well, into the analysis, chapter nine pays particular attention to examining the available evidence on intra-Asian and extra-Asian interdependencies. Chapter 10: Conclusions on the architecture of economies. The concluding chapter reviews the evidential strengths and weaknesses of the cluster complexes framework and evaluates the use of input-output data. From the analysis of the spatial structure of linkages, it is apparent that there are differences in the way countries and industries are connected. Some national clusters are hubs for a large system of networks, with some countries heavily dependent on goods developed in other countries. There also appear to be differences in the global organisation of assembler industries (e.g. motor vehicles) and those with modular product architectures (electronics). The chapter also emphasises future research possibilities.

1.4.3

Boundaries and Contributions of this Research

This research project behind this book could not possibly address all of the important questions related to the structure of linkages beyond the borders of regions. For example, while, the systems of innovation literature emphasises the historical, institutional and government policy factors in promoting system innovativeness, there is not the space here to consider in depth how these shape cluster network structures. In addition, the structure of intra-national linkages has been left to a future date. However, this book contributes to the analysis of innovation and the modelling of inter-regional interactions, in a number of ways. 1. For innovation systems research: • This book identifies a serious deficiency in the way it currently treats industrial production and geography, suggesting that multi-spatial analysis should be included as one aspect of the research agenda; and • This book further reveals that the current approach to ‘national’ or local systems is ignoring the scale and spatial structure of the transfer of components across borders and the various types of connections that exist between production locations.

1.5

Overview

11

2. In the field of spatial modelling the present book: • Applies a proposed methodology (Cooper 2000) for calculating net multipliers to a series of large multi-country datasets to analyse the linkages between industries and countries across time and at various levels of detail; and • Reports on the development of various new measures of the strength of linkages between economies which generates new analyses of industrial interdependencies. While, the data analysis reported here remains structural, an effort has been made to interpret them within the wider technological, organizational and innovation systems contexts.

1.5

Overview

It is obvious that in moving away from simply locating clusters and assessing their advantages to a framework that examines them within longer value structures, the issue of the treatment of political borders with the innovation systems literature is an important sub-theme. As the dust jacket for a recent book on clusters reads ‘… the authors are able to explore the role that national innovation systems play as a framework in which clusters operate’. It was, therefore, felt that to construct an innovation system framework around value chains, it was essential that the study needed to include a careful consideration of why borders are so important in the study of innovation. It is not the argument here that national borders or regional systems do not matter, clearly, they remain important (see in particular the analysis of the role of borders on trade in Sect. 4.2.2). It is, however, the thesis of this book that there is a need for an extension of the vision of the innovation systems approach to go beyond borders and examine how the different systems interact. The corollary argument is that it would be equally mistaken to move from the current situation to a position where international links are only generally considered. In this way, the current project goes some way towards addressing an earlier proposal by Lundvall, when he suggested: ‘It would be very interesting to extend the mapping of innovation flows to international input-output tables, in order to see what degree flows of innovations are international rather than national, and how this differs between sectors. Again this would present us with very complicated methodological problems. In order to understand the international relationships, it might be necessary to combine the structural analysis with an institutional approach’ (1996, p. 363).

The connections between systems need to be understood in terms of the role, scale and spatial structure of those links – not as local or global but as structured networks. Although this study has addressed the issue of internationalisation, the findings are very likely applicable intra-nationally.

12

1

Introduction

Finally, for the reader, the goal of the empirical analysis of the current work has been to reveal the emerging shape of global production and innovation based on analyses of trends across a 30 year period. Inevitably, new databases and analysis are constantly under-development, it is neither possible to predict what these might show or to predict the extent to which changes now underway (in countries such as China for example) will impact on the spatial structures of production that are examined here. However, the system characteristics described in this book (specialisation reinforcing simultaneous clustering and fragmentation) does appear to be a robust feature of the organisation of economies.

Chapter 2

Systemic Innovation and Nation States

2.1

Knowledge, Technology, Innovation and Nation States

The decade of the 1980s was one where Japanese industrial competitiveness in a number of technology based industries notably computers, consumer electronics and motor vehicles came to the fore. Dosi et al. (1990) considered this example of Japanese development as the only case of economic catch-up from less developed to advanced economy status in the post World War II era. Throughout this catch-up period, corporations based in Japan expanded their technological capabilities and progressively introduced new processes and products that were at the leading edge in many industries. With the exception of the case of Japan there was, at least until the mid 1990s, an interesting feature of research and development (R&D) indicators in particular, and innovation indicators more generally. Relevant data reveals that the relative position of many OECD countries does not vary significantly across time (Patel and Pavitt 1994). As one example, Voyer (1999) argues that Canada did not significantly increase its gross expenditure on research and development (GERD) across a 30-year period. These observations – one country’s improved competitive position (Japan), the inability of Canada to improve its expenditure on research and development and the general pattern of the difficulty for countries to shift their trajectories – strongly suggests that a company’s technological capabilities are not just its own business. The progress of the major Japanese companies, in particular, had a considerable impact on researchers with an interest in technological competitiveness. The success of Japanese organisations suggested that the national milieu in which businesses were embedded is influential in their behaviour and success. As this evidence came to light in the 1980s, the concept of national innovation systems (NIS) emerged (Freeman 1987 and many since). Two decades later, this concept is discussed or implied in the vast majority of research articles on innovation. NIS research has revealed that a country’s research and education systems, its government policies on industry and innovation, and even the operation of its labour market, can all affect the ability of a firm to introduce new products and processes. Access to highly skilled personnel, appropriate infrastructure and an environment conducive to co-operatively developing research opportunities between publicly funded research laboratories and business, all play a part in the overall innovativeness of economies.

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_2, © Springer-Verlag Berlin Heidelberg 2009

13

14

2

Systemic Innovation and Nation States

Fig. 2.1 National innovation systems in the literature

This chapter analyses the major threads of NIS research, as it is relevant to international production structures – which is not by any means the totality of the NIS literature. It evaluates the role of the ‘system’ in shaping technological competitiveness, promoting economic growth and the way in which knowledge production remain remarkably stabile through time. Figure 2.1 locates this chapter’s contents within the overall research framework of the book. However, the purpose of this chapter is to reveal how analysis of these subjects has been carried out largely by treating each nation essentially as independent entities. National innovation systems research has focused on determining the characteristics of particular countries that have aided or hindered economic success. As this chapter reveals, there has been little work on the linkage between the knowledge economy and the national economic interdependencies within the growing structure of global production. This chapter also shows the NIS focus on the importance of the internal dynamics of economies, to the neglect of external connections, can be seen in the very early work of Josef Schumpeter.

2.2 2.2.1

Technology, Competitiveness and Systems Economic Growth

In the long run, what matters for human societies are economic growth (World Bank 1999), environmental sustainability (Diamond 1998 and World Commission on Environment and Development 1987) and social capital (Vinson 2004).1 Although 1 A revealing study of the importance of ‘social cohesion’ in economically disadvantaged communities in rural, regional and urban Australia. It reveals that outcomes for communities with greater social capital were better than for those with less social capital.

2.2

Technology, Competitiveness and Systems

15

this triple bottom-line has been gaining more prominence in recent years, the economic wealth of nations is critical for quality of life and the health of populations. As the history of economic development is such a popular topic (Landes 1999), it is a little surprising that the understanding of the conditions that facilitate economic growth is advancing very slowly. Temple (1999, p. 112) goes so far as to say that the analysis of economic growth has often been a backwater within macroeconomics. This neglect makes it is easy to make big claims for what is known of economic growth, particularly as it relates to the role of new technology, but unfortunately less seems to be known than is claimed. From the early 1960s, researchers from the neo-Schumpeterian tradition have been interested in the role of new knowledge and innovation in economic growth. More recently, within the neoclassical economics tradition ‘new growth theory’ emerged with the work of Romer (1986) who began to econometrically model the role of new knowledge creation in fostering economic growth. The OECD (2001b and OECD 2003a) has recently summarised the evidence on the empirics of economic growth. Neo-Schumpeterian authors have approached the analysis of economic growth from a number of angles. Fagerberg points out that prior to new growth theory it was expected that capital and labour contributions would be the preponderance of factors contributing to economic growth (2001). As much as 80% of growth (p5) remained outside the models in the first attempts at growth accounting. Fagerberg argues that there are various bits of evidence supporting the significance of technology diffusion, knowledge spillovers and localisation of knowledge in economic growth. However, Temple (1999), in analysing the empirical evidence on these and other factors, is less confident that there are any clear answers. Importantly, there is a difference between national economic growth that can be influenced by many factors, such as macroeconomic management and consumer confidence related factors, and business competitiveness, which is more directly related to innovation. The OECD highlights the growth experience of Member countries and in so doing reveals the difficulties of determining the factors that encourage economic growth. Some point to the role of new technology and innovation, but if that were the only answer, then why did growth languish in Japan, which has a large and successful computer hardware industry, but soar in Australia, which has virtually no such sector at all? (2001b, p. 9)

The OECD’s conclusion, while hard to fault, does not actually help very much. ‘Consequently, policies that engage ICT, human capital, innovation and entrepreneurship in the growth process, alongside fundamental policies to control inflation and instil competition, while controlling public finances are likely to bear the most fruit over the longer term’ (2001b, p. 10).

Although new technologies play a part in national economic growth, determining the value of that contribution remains a challenge. Partly, the problem with calculating the outcomes of aggregate investment in new technologies is the urgent need for improved theoretical models and tools for measuring creative destruction processes, as noted by Haltiwanger (2000). New technologies

16

2

Systemic Innovation and Nation States

do in time replace2 older ones or change the relative pace at which different sectors change rates of employment or productivity. As global competition increases, the race to develop new technologies may just help countries keep pace with one another. Since the 1970s, there have been some remarkable changes in the world economy. New technologies and innovation have become ever more important to business success. The massive increase in computing power with benefits for national productivity (Jorgenson et al. 2003), the Internet and with it the emerging possibilities of electronic commerce (OECD 1999a), the beginnings of biotechnology for both health and agriculture (Industry Canada 1998) have encouraged governments to focus on the economic potential of emerging technologies.3 Nanotechnology is already being seen as the next ‘big thing’, with substantial investment in the USA (through the National Science Foundation – NSF). The European Union (EU) is also focused on this set of technologies, already devoting a chapter to measuring the level of investment in Member countries in the European Commission’s 2003 indicators report. Business communities now lobby Government for innovation programs (see, e.g. Australian Industry Group 2002) and technology and knowledge is now an issue for debate by neo-classical economists. Even a bastion of economic theory and tradition, The Economist journal reflects these changes with a quarterly review of science and technology. Perhaps most surprising, governments in the world’s poorest countries, such as Mozambique, have a growing interest in science, technology and business innovation (Garrett-Jones et al. 2003).

2.2.2

Innovation Aids All Sectors from Agriculture to Services

Academic interest in industrial innovation is often traced back to Joseph Schumpeter, an Austrian economist, who until recently received little attention in economics courses. In ‘A Theory of Economic Development’, first published in English in 1934, Schumpeter (1968) considered technological innovation not merely as the driving force of growth but as being the very essence of development (p65 ff). Schumpeter’s insights are now largely supported by the results coming from a broad range of authors. Porter (1990) suggests that the ability to create new products or processes allows firms to be freed from the traditional sources of advantages such as lower labour costs and resource access, to be replaced by the advantages of skilled labour and technological infrastructure. But this change from ‘comparative’ to ‘competitive advantages’ does not make the

2

See Christensen, Craig, and Hart (2001). The USA’s Government expenditure on health research alone has grown from around US$15b in 1998 to nearly US$30b in 2004 (constant 2004 dollars) AAAS (2004b). 3

2.2

Technology, Competitiveness and Systems

17

factors behind business development less local – it makes some factors even more localised. By the 1960s and 1970s, Schumpeter’s ideas on the innovation process were being put to the test with studies of company research and development practices and their capacity to introduce new products to the marketplace. Freeman comments on one of the path-breaking studies of the time: ‘The SAPPHO project (Freeman, 1974 and Rothwell 1974) had already shown that good internal coupling between design, development, production and marketing functions was one of the decisive conditions for successful innovation. Many failures could be attributed to the lack of communication between the R&D, production and marketing functions as was also shown in the brilliant sociological study of Burns and Stalker (1961)’ (1994, p. 472).

These early studies focussed primarily on the development of new products that were based on scientific research by firms in the manufacturing sector. While, this focus inevitably concentrated analysis on industries that require more research and development (electronics, pharmaceuticals, aerospace and motor vehicles), over time this has changed. Nelson (1993b, p. 513), for example, in his volume comparing national innovation systems demonstrates that there is a strong connection between a nation’s competitive agricultural sector and the funding of agricultural research. It has taken much longer for service businesses to be treated on their own merits, in terms of innovative capacity. Initially, services were understood for their role within the traditional areas of interest (manufacturing and new technologies). Freeman noting the significant investment by business service firms in technological change, comments: ‘In house software development… is now characteristic of many firms in financial services, who also have a heavier investment in ICT equipment than most firms in manufacturing. At the same time, specialist software companies are proliferating and have a very dynamic role in technical change’ (1994, p. 478).

There continues to be interest in how services and manufacturing interact in the innovation process (Tomlinson 1997) and there is a growing body of research on the dynamics of services innovation (see Tether and Metcalfe 2003 and Baark 2001). The SI4S4 and RISE5 projects have greatly improved the information on services innovation and the role of public sector research organisations. This broadening interest of innovation researchers now even extends as far as consumer preferences with a research project on ‘consumption and demand’ (Harvey et al. 2001). Innovation, is not just the activity of high technology manufacturing firms, it is the business of firms throughout a modern economy. Nevertheless, the important shift from the early 1970s has been the change in emphasis from the characteristics of firms to the systematic properties of innovation, typically characterised at the level of the nation state. 4 5

http://www.step.no/old/Projectarea/si4s/start.htm accessed 4 Dec 2007. http://centrim.bus.brighton.ac.uk/open/we/do/proj/rise/index.htm accessed 4 Dec 2007.

18

2.2.3

2

Systemic Innovation and Nation States

Innovation Embedded in Nations

In neo-classical trade theory, the standard factors of production are traditionally considered as national but knowledge is considered to be freely available (Lundvall 1998). This should lead to the global movement of production to where there are comparative advantages. However, neo-Schumpeterians have a better understanding of knowledge, highlighting that it is not easily accessible and is related to industrial structure. Therefore, they continue to retain a focus on the role of the nation state. While there are commonalities within the definitions of what is systemic in innovation processes within nation states, there is no agreed theory of cause and effect mechanisms. In Edquist’s (1997b) view, systems of innovation, which encompasses a range of spatial scales (not just nation states), is a framework for investigating the development and evolution of technological capabilities, concentrations and specialisations. Edquist provides a guide to the systems of innovation perspective, emphasising it: • Is ‘holistic and interdisciplinary’ – for constructing a broad understanding of the ‘determinants of innovation’ (1997b, p. 17); • Is a presentation of the ‘historical perspective’ on geography and natural resources access, etc (1997b, p. 19); • Focuses on the ‘differences between systems and non-optimality’ (1997b, p. 19) – all systems are different and defining a priori an optimal system is not sensible; • Stresses ‘interdependence and non-linearity’ (1997b, p. 20) – as it is ‘an approach in which interdependence and interaction between the elements in the system is one of the most important characteristics’ (1997b, p. 21); • Focuses on ‘product technologies and organisational innovations’ (1997b, p. 22); and • Places institutions at the centre of analysis. This list by Edquist breaks down into essentially two categories: the first three points position the approach to the economics of innovation and technological change vis-à-vis neo-classical economics; while the second three points identify the key issues in the study of innovation. Institutions, interdependencies and product innovations in a particular place draw upon history, culture and a set of policy interventions. This implies that the nation state is seen as a natural boundary for the actors and activities that are relevant to the creation of economically useful knowledge. Freeman (1995) reviews a number of cases where the different way innovation systems developed was important to the broader trajectories of economic development. He argues that Japan encouraged the integration of production with research and development and technology acquisition while in the USSR these were all components that were separated from one another. In Japan, networks of users and producers developed while in the USSR these never developed. In East Asia, there was heavy investment in the education systems but in Latin America, the education systems deteriorated. In East Asia, industrial research and development expenditure increased, however, in Latin America it remained steady.

2.2

Technology, Competitiveness and Systems

19

The emphasis on the nation dates back at least as far as Schumpeter. A chapter written by Schumpeter for the first edition of his ‘A Theory of Economic Development’ was not included in the first English translation has, comparatively recently, been translated into English and analysed. The text indicates that back in the early years of the twentieth Century Schumpeter already envisioned economic growth deriving from entrepreneurship as an endogenous process captured by nations. Matthews writes: ‘It is this seventh chapter, lost to the world after Schumpeter’s decision to drop it from his second edition (which then formed the basis of the English translation published only in 1934) […] The chapter, entitled Das Gesamthild der volkswirtschaft (the economy as a whole) provides a fascinating missing “chapter” in Schumpeter’s thought, previously inaccessible to the English-speaking world. The chapter clearly written in haste late in 1911 to catch a printing deadline, sketches a highly original summation of his model of internal economic development, where transformation is generated from internal dynamics represented by entrepreneurial initiative – in contrast with the prevailing doctrines which saw change in economic circumstances, and growth, as responding to external stimuli, such as population growth, or technological innovation, or the opening up of new geographic markets’ (2002, p. 2).

This emphasis on the role of domestic dynamics contrasts starkly with the literature and policy recommendations that emphasise export-oriented growth and the contribution of global economic growth to national growth performance. The IMF (2001) has had a consistent focus on encouraging developing and heavily indebted countries to support export oriented industries. In this view, continuing liberalisation of national economies will lead inevitably to a decline in the importance of the economic and social policy making of national governments. As Rangan and Lawrence argue: ‘In the absence of border barriers, competition would be global. Corporations would rapidly shift to locations that offered lower costs. Indeed, global competition would compel them to do so, because victory would go to the firms with the lowest costs, whereas firms mired in high-cost locations would eventually be driven out of business’ (1999, p. 4).

On the other hand, Schumpeter’s view that domestic capabilities matter is supported by the empirical evidence presented by Rowthorn and Kozul-Wright in their analysis of globalisation. ‘Domestic determinants of economic growth remain significant’, even though capital investment flows are increasing, because the drivers of ‘capital accumulation retain domestic roots’ (1998, p. 31). Thus, those that claim globalisation is the death of the economic influence of nation states appear to have rushed to judgement. The OECD (1992) suggests that the effective structuring of a country’s NIS can help a country progress rapidly and that conversely weaknesses may lead to the squandering of other resources.

2.2.4

Systems Theory

Although it took some time for implications of innovation based competitiveness theories to be applied to the full range of industries, the change to systems thinking, by comparison, was relatively quick (a point noted by Edquist (1997b, p. 3). In the

20

2

Systemic Innovation and Nation States

late 1980s, the concept that innovation was a systemic property of the nation state emerged with Freeman’s (1987) use of the term ‘National Innovation System’ (NIS) with an application to the growing strength of Japanese businesses. Since then there have been many systems of innovation approaches (regional, technological, sectoral), but the national perspective remains the primary framework. The purpose of this section is to present an overview of the major strands of NIS theory and evidence as they relate to the structure and evolution of the geography of global production. The international circumstances which encouraged the neo-Schumpeterians to analyse the role of the nation state led others to a similar interest and reinvigorated debates about the future of nations. Tyson (1992) perceived economic policy in national terms, whilst Ohmae (1995) was predicting the looming irrelevance of the nation state. Through this period of the late 1980s and the first half of the 1990s, research on economic growth returned as a core topic in economics (see for example Mankiw 1995, and the reviews in The Economist 1992a, b; 1996a, b). The nation state has come to be at the centre of innovation studies because the ability of companies’ to develop new products, processes, services and technologies does not rest solely on the resources and capabilities that can be controlled by the firm itself. NIS is a way of considering variables that are both within the influence of nation states and those which, although difficult to change are nonetheless apart of the evolutionary patterns of country development. In the former category are government interventions such as; industry policy, the higher education system, technical education, social welfare and public R&D funding (level and research fields). In the latter category are specific features of an economy such as specialised supplier businesses in particular industries, venture capital access and the way labour relates to employers (pay and non-pay conditions), all trajectories that are difficult to influence. What does define an NIS? There are common themes but there are many subtly different emphases in the various definitions. The first books on this topic Lundvall (1992a) and Nelson (1993a) took quite different approaches. Nelson’s book was a country-by-country description that was prone to the individual predispositions of the local authors. For example, the author of the chapter on Australia (Gregory 1993), a labour market economist, emphasised the role of human resources and the organisation of labour relations for innovation capacity. Lundvall’s book, in contrast, explored the theoretical dimensions and presented empirical data on innovativeness and technological competitiveness within a thematic presentation. Lundvall remains a key promoter of the NIS perspective and the nationalism of innovation (1998, with Maskell 2000 and et al. 2002). In trying to define NIS, Freeman suggests that it is more than just R&D but encompasses the entire way a nation is organised, commenting: ‘most neo Schumpeterians, following Lundvall (1992) and his colleagues, stress that a ‘national system of innovation’ is much more than a network of institutions supporting R&D, it involves inter firm network relationships and especially user-producer linkages of all kinds (Anderson 1992a) as well as incentive and appropriability systems, labour relations and a wide range of government institutions and policies’ (1994, p. 484).

2.2

Technology, Competitiveness and Systems

21

Smith, in contrast, places more emphasis on the interactions between different elements of the knowledge production system (companies, labs, etc) and the facilitating institutions (the legal system). Smith also emphasises the influence of cultural factors (values and norms), arguing: ‘the innovative performance of an economy depends not only on how the individual institutions (e.g. firms, research institutes, universities) perform in isolation, but on how they interact with each other as elements of a collective system of knowledge creation and use, and on their interplay with social institutions (such as values, norms, legal frameworks)’ (Smith 1994, p. 3).

Another definition has a greater focus on the role of government, Metcalfe suggesting: ‘A national system of innovation is that set of distinct institutions which jointly and individually contribute to the development and diffusion of new technologies and which provide the framework within which governments form and implement policies to influence the innovation process. As such, it is a system of interconnected institutions to create, store and transfer the knowledge, skills and artefacts which define new technologies’ (Metcalfe 1995, p. 463).

The OECD (1999c) combines many of the variables in these and other definitions of NIS in a diagram of the operations of a national system, but neither the diagram nor the book, in which it appeared, goes beyond generalities. However, The OECD is not alone in attempting to create ‘maps’ of the innovation system. The Australian Government has recently published its guide to Australian innovation (Science and Innovation Mapping Taskforce 2003). The report states that ‘For the first time, we have been able to present a detailed overview of our science and innovation system in Australia’ (2003, p. i), but it presents little that is different from other indicator publications that have been published in Australia for sometime (see for example Department of Industry, Technology and Commerce 1987). Furthermore, its analysis is commonplace in other countries (National Science Board 2002 and European Commission 2003). What becomes clear, however, from these attempts at defining and mapping innovation systems is that we are far from a macro-innovation theory of economies, in the same way that macro-economics purports to depict the interactions that occur within modern economies. It may not even be desirable to progress towards such a theory. Instead, NIS is a body of research which adopts a number of different viewpoints and methodologies which nevertheless all progress the general proposition that political geographies still matter for the innovativeness of businesses located within their borders. Dosi (1999) provides a much needed classification system of the various research approaches to analysing the influences on innovation which are ‘national’. Dosi categorises both the major processes driving systems of innovation as well as the diverse meta views on the operational features of national systems (with appropriate author attribution). Processes in national innovation systems: 1. Production systems. 2. Innovation system operations. 3. Knowledge accumulation.

22

2

Systemic Innovation and Nation States

Dosi’s typology of analytical lenses on national significance for innovativeness: 1. National innovation institutions & policies [R&D funding and universities, etc] (Nelson) 2. User-producer relations (Lundvall) 3. Technological accumulation (Patel & Pavitt) 4. National institutions [Financial markets, labour markets and training institutions] (Soskice) From these two lists, three categories are important for the analysis in this book. 1. Technological accumulation (see Sect. 2.5 of this chapter). Knowledge and technology creation are viewed as important engines of long-term economic growth (Nelson 1990) but, like natural resources, they are accumulated unevenly across nations. Similar to other resources, history plays a part in exploitation and the rate at which countries develop knowledge resources is fundamentally shaped by the operation of national institutions. Knowledge typically accumulates along pathways. Investment in science and technological capabilities in one generation is usually consistent with previous investment decisions. This is because knowledge production relies on self-reinforcing mechanisms including physical capital (infrastructure in the form of existing laboratories or expensive scientific equipment) and human capital (training of personnel). Knowledge also tends to be ‘sticky’ (Dosi 1999) to given locations both because it is accumulating and as most knowledge is uncodified – it does not diffuse easily. Such tacit knowledge is held in the minds of people and is passed on by either learning by doing or word of mouth. Increased knowledge accumulation is thus likely to foster increased diversity rather than convergence amongst the world’s economies and is one of the prime reasons for any understanding of globalisation retaining a focus on specific locations. 2. Production systems (see Sect. 2.3.2 of this chapter). Systems of innovation research contributes to improving the understanding of production specialisations, trajectories and industrial location at the national and the sub-national level.6 NIS analyses have focussed on examinations of industry competitiveness (Fagerberg 1998), business R&D (Patel and Pavitt 2000 and Gassman and von Zedtwitz 1998), patenting patterns (Archibugi and Pianta 1992) and export growth (Dalum 1992 and Laursen 1998a, b). Although the framework of innovation systems extends beyond industrial activity, it is necessary to focus on corporations, industries, sectors and whole value chains. It is through these actors and activities that technologies come to market. 3. User-producer relations (see Chaps. 3 and 5). One of the most important findings of innovation research is that the interaction of the producers and users of industrial

6 See chapter 3 for a discussion of geography based analyses including; development blocks, regions and clusters.

2.3

The Historical and Continuing Importance of Nations

23

components and services can create an environment for new products to emerge. The communication of needs (users) and possibilities (producers) opens the way to the creative processes and the incentives for investing in research and/or product development. Thus, ‘user-producer relations’ are seen as one of the key characteristics of business innovation (von Hippel 1988, DeBresson 1996 and Edquist 1997b). In some cases the intensity and nature of relations has been seen as an attribute of nation states (Lundvall 1992a). The cooperation seen in user-producer relations runs counter to the assumptions of market competition and arms length open contracting that exists in neo-classical economics. Far from simply emerging from the operation of markets, competition and inventive businesses, the development of technology evolves from an interaction between the different actors (see Hofer and Polt 1998) and can even be observed at the system level (DeBresson 1996). Unfortunately, there has been very little analysis of cross border user-producer relationships. Although DeBresson et al. (1998) imply a positive role for international linkages; they could not provide information on the scale or spatial structure of the extra-territorial links. The issue of linkages between businesses across borders is quite different to the topic of international R&D spillovers which has been of some interest to researchers but which have been perceived as small (i.e. see van Pottelsberghe 1998). Due to its relevance for developing an understanding of the extension of innovation systems across borders, the topic of user-producer relations research is returned to throughout this thesis. The next two sections further develop the reasons for considering nations as the appropriate spatial scale at which to analyse innovative activities.

2.3

The Historical and Continuing Importance of Nations

Neo-Schumpeterians, to date, have seen competitiveness predominantly in national terms. It is thus necessary to map this existing thinking on the role of nations before being able to highlight the weaknesses of the approach and the need to look at the interdependencies between systems.

2.3.1

Learning Nations

The ability of countries, through their businesses and populations, to learn, to access current knowledge and to understand how to transform existing data into new knowledge is crucial to social and economic progress. Although nations have developed many traditions in the funding and organisation of their education systems, some features are arguably fundamentally ‘national’. Qualification standards are often a national responsibility and intra-national labour mobility is rarely hindered, whereas movement across international borders is more restricted.

24

2

Systemic Innovation and Nation States

Accessing human capital and knowledge have been critically important for economic success since at least the beginning of the industrial revolution, even earlier according to Smith (2000). In the view of Landes (1999) access to codified knowledge is no guarantee of being able to successfully adopt technologies, noting that during the industrial revolution even with ‘sample products and equipment’ or ‘blueprints and explicit instructions, some know-how can be learned only by experience’ (1998, p. 278). Not surprisingly, European countries designed policies, during the industrial revolution, that not only built their own capability but aimed at acquiring new ones through, initially, hiring British workers; ‘foreign governments paid people to come and helped them set up in business’ (Landes 1999, p. 279). Landes comments that some of these workers were: anonymously ordinary, most British expatriates were workmen drawn by wages that ran twice and three times higher than home. (British wages were ordinarily considerably higher than those across the channel, but these experienced craftsman and mechanics were scarce commodities in follower countries) (1998, p. 280).

Whilst the measures taken by follower countries were partially successful, it was the eventual development of formal technical education systems, first in France and then copied across Europe that powered success. Germany developed the approach to its fullest extent at that time. It formed a network of trade schools and technical high schools as well as changing the universities to conduct teaching and research in chemistry and engineering (Landes 1999, p. 283). The German universities became centres of technological diffusion because they focussed on both theory and applications of science. In contrast, the approach in Britain still relied upon ‘learning by doing – the strategy had driven the Industrial Revolution’ but failed as ‘the frontiers of technological possibility and inquiry moved outward, exploration went beyond the lessons of sensory experience’ (1999, p. 283). The education system helped Germany take the lead in the chemicals industry, in which Britain had previously had an obvious competitive and comparative advantage (Landes 1999). The change in the structure and content of the education system in Germany led directly to significant changes in the economic fortunes of nations and accords with Lundvall’s strong emphasis on the learning capabilities of nations. He notes ‘Innovation appears now, not primarily as a single event, but rather as a process’ (1992b, p. 9). Thus it is: argue[d] that most important forms of learning may fundamentally be regarded as interactive processes, and that together with the economic structure and the institutional set up form the framework for, and strongly affect processes of interactive learning, sometimes resulting in innovations (1992b, p. 9).

For Lundvall a national framework is necessary for understanding this learning process for the development and diffusion of knowledge, which is critical to the creation of innovations and competitiveness (see also 1998, with Maskell 2000 and et al. 2001). Consistent with this perspective, are the conclusions that universities and publicly funded research needs to be seen primarily as an investment in the development of human capital rather than investment in new technology

2.3

The Historical and Continuing Importance of Nations

25

(Salter et al. 2000). Although new technologies may or may not emerge, it is talent not technology,7 which has the greater economic benefit. The investment payoffs include (amongst others) ‘increasing the stock of useful knowledge’, ‘training skilled graduates’ and ‘creating new scientific instrumentation and methodologies’ (2000, p. 59).

2.3.1.1

Knowledge Accumulation

Increasing knowledge creation in a given field tends to be dependent on previous investment in knowledge creation. The idea of cumulative causation in traditional economics has been discussed for the better part of a century (see Toner 1999) and its application to endogenous growth theory has been intensively debated over the last decade. The emphasis historically has been on increasing economies of scale, increasing specialisation and the central importance of manufacturing. It is, however, the cumulative causation of investment in knowledge that has gained very wide acceptance with the neo-Schumpeterian community. Investment in knowledge is path dependent and generates positive feed back effects. Capability and capacity with one field of knowledge enables future research within that same field but it is difficult to change scientific fields. Each field of knowledge requires existing training expertise, scientific infrastructure and relationships with researchers across the world facilitating access to tacit and codified knowledge. Cohen and Levinthal (1990) were the first to suggest that there are ‘two faces of R&D’. Business expenditure on research and development both generates new knowledge and, more importantly, it assists the company in attempts to access the greater part of knowledge that is external to the business. Without conducting its own R&D, companies will be largely ignorant of the leading edge and be unable to absorb new technologies. Cohen and Levinthal also argue ‘accumulating absorptive capacity in one period will permit its more efficient accumulation in the next’ (1990, p. 136). Dowrick, quoted by the Industry Commission (1995) in its wide-ranging inquiry into the importance of R&D to Australia’s competitiveness and the role of R&D policy, notes that R&D investment also has positive feedbacks and spillovers: Crucially, the larger the stock of knowledge, the easier it is to increase it. Better educated and more knowledgeable people learn faster and develop new ideas more easily’ … ‘The second channel is R&D spillovers, which involves the notion of transfers of knowledge among firms for which no payment is made (1995, p. 153).

There are many ways of improving an organisation’s knowledge base including research, learning by doing, reverse engineering, imitation and the purchase of equipment (Dosi and Castaldi 2002) but all encourage specialisation and trajectories. Patel and Pavitt (1998) argue that technology development by nations is both

7

This is also the title of their report.

26

2

Systemic Innovation and Nation States

uneven and divergent and earlier (1994) they suggested that while national technological specialisation trajectories are not predetermined, the incentives structure and the available resources do create likely evolutionary patterns. Dosi and Castaldi (2002) find evidence in the literature to support the argument that, globally, countries have diverging technological capabilities. However, this trend is less pronounced for countries within the OECD group. Interestingly, economic growth patterns reveal that the advanced economies are converging with each other but largely diverging from the rest of the world (see, e.g. Dowrick and DeLong 2001).

2.3.1.2

The Non-Globalisation of Corporate R&D

Knowledge (as opposed to information) is not only cumulative, but is also geographically ‘sticky’ (in Dosi’s 1999 words). Such stickiness derives from several dimensions including the cumulative nature of knowledge specialisations (as above) and the limited geographic spillover of tacit knowledge (discussed below in Sect. 3.4.38). In terms of the role of nations, it might be expected that as production activities are internationalised, multinational corporations may wish to internationalise their research efforts. Neo-Schumpeterian research has, however, not found strong evidence for the decentralisation of R&D centres away from being geographically near the home bases of multinational corporations. Patel (1997) found that firms tend to keep their R&D activities in their home country, although there is a trend towards increasing the amount of R&D that is located abroad (see for example OECD 2008b). While generally supporting this existing view of the involvement of multinationals in R&D, Carlsson suggests that ‘innovation systems may have become more leaky over time. The role of tacit knowledge and the spatial limits of knowledge spillovers have caused firms to locate R&D facilities where new knowledge is being created’ (2003, p. 21). There is a significant literature on the knowledge production and absorption-activities of multi-national enterprises9 and international knowledge spillovers.10 This line of research is not pursued in depth in the present thesis, as the purpose here is not to map the extent to which knowledge flows, but to begin to develop an understanding of how local knowledge contributes to local specialisations within internationally extended value chains.

8 In chapter 3 (below) it is shown that the limited spatial spillover of tacit knowledge is a key argument for industrial agglomeration. 9 Readers interested in R&D location and internationalisation can start with Paoli and Guercini (1997), Gassman, and von Zedtwitz (1998a) and OECD (1998a). 10 On the role of FDI and international knowledge spillovers – see Van Pottelsberge De La Potterie (1998) and Verspagen and Schoenmakers (2000) on the spillover of knowledge in patents.

2.3

The Historical and Continuing Importance of Nations

2.3.2

27

Production System Specialisation and Trajectories

As noted above knowledge specialisations accumulate in trajectories and it can also be shown that knowledge and industrial specialisations trajectories co-evolve. It has already been pointed out that in the early stages of the industrial revolution the development of formal training systems aided Germany’s ability to gain the predominant share of the European industrial chemicals sector. Landes describes this shift as one of the most rapid industrial transitions in history (1998). This link between learning and industry is not, however, just an interesting facet of the birth of the modern age. A statistically significant association between scientific performance and economic specialisation for science based, scale intensive11 and some resource-based industries has been found by Laursen and Salter (2001, p. 18). With this connection in mind, Wixted (2005, p. 35) charts the national configuration of manufacturing value added and manufacturing exports (2005, p. 36) for OECD countries for 1994. The first chart reveals that Germany, Japan and Korea are countries with a strong manufacturing presence in a number of industries. In contrast to the first chart, the second demonstrates that many countries have export specialisations in the same industries. The transport, non-electrical (industrial) machinery, electrical machinery as well as food and textiles all appear as having overseas sales above 2% of GDP for a number of countries. As most East Asian economies are not represented in the database, export strengths in electronics do not feature prominently. This cross-country profile of export specialisation is consistent with the World Trade Organization (2003) list (in order) of the most globally traded industries; transport and machinery equipment, office and telecom equipment, mining commodities, chemicals, automotive and agricultural products. How these country specialisation patterns emerge and how they evolve has been of interest to neo-Schumpeterians, although the processes leading to increased levels of cross border economic activity in these industries is of less interest. Evidence on structural stability of the specialisation of economies is a key piece of evidence in debates over the nationally bound conditions for knowledge creation and innovation which are the basis for arguing for NIS. In particular, the speed of change can be used as a measure of the scale of economic movement and the continuing importance of both the nation state and knowledge production for industry. Slow change, for example, is used as evidence that ‘globalisation’ processes are limited by endogenous factors of knowledge accumulation. The evidence, in fact, is that industrial structures do evolve relatively slowly. Archibugi and Michie (1998) note that manufacturing specialisations change ‘very slowly’ with it being difficult to ‘move from an established competitive advantage in one industry to another’ (1998, p. 11). These authors draw attention to the underlying

11 These classifications of industries are also used in the current analysis and discussed in chapter six, and the link between science and scale based activities is interesting in the light of the research presented there.

28

2

Systemic Innovation and Nation States

technological competencies of industries, linking industry development to the cumulative processes of knowledge progress. Industrial employment structure is shown by Metcalf et al. (2002) to change, at a fairly constant rate, but industrial structure as measured through production output shares exhibit a degree of continuity over approximately 20 years. Output shares appear to resist change for a period of time and then change can occur quite rapidly – relative to the initial conditions. At the level of manufacturing sector sub-branches, Wolff argues that the industrial specialisation of OECD countries changes very slowly, commenting: ‘The finding of little change in the degree of specialization among manufacturing industries may appear somewhat surprising in light of the evidence that aggregate measures of factor endowment (such as capital-labour ratio for the whole economy) have become similar in these advanced economies. On the other hand, the result is consistent with the finding that dispersion of productivity at the industry level remains high, and that there has been no strong trend toward cross-country convergence of industry-level productivity since mid 1970s. It appears countries are maintaining specializations in different industries; in this way convergence of aggregate productivity can be consistent with continuing divergence of industry-level productivity and a continuing high dispersion in production patterns’ (2000, p. 200).

Curiously, therefore, despite a convergence of factor endowments across advanced economies, industry specialisations are not converging. Even so-called low technology industries are resistant to wholesale movement. Dosi et al. (1994) suggest that textiles and clothing have often been the starting rung for industrialisation for developing economies and, by implication, is more easily internationalised. However, low technology industries have, in a number of cases, not moved offshore as expected. In an article titled ‘the strange life of low tech America’ (1998, pp. 81–82), The Economist, in almost surprised tones, explores the continuing success of some very low cost, low technology activities in the USA. Amongst the various explanations of this success are the prevalence of trade protection barriers, the benefits of being close to markets for these industries, and the accumulation of skills which are necessary for industry competitiveness together with unexpected levels of ongoing product development in these industries. Such trends point to the importance of local sources knowledge and innovation, and it is argued by some that these sources are likely to remain predominantly national without large movements across the borders of developed countries (see Ernst 2000). This ‘spatial stickiness’ (Ernst 2000, p. 2) of knowledge and innovation facilitates countries developing measures that maintain their technological superiority and could be behind the increased movement of goods and services, as there is a growing need to integrate technologies that have not been developed ‘in country’.

2.4

Nations in a Changing World: Weaknesses of NIS

The evidence presented so far in this chapter provides strong grounds for continuing to believe that political nation states remain economically important domains of the world economy. Particular characteristics of the processes that lead to the generation

2.4

Nations in a Changing World: Weaknesses of NIS

29

of new knowledge, together with the dynamics of technology diffusion, appear to preference national spaces over a borderless world. This does not, however, imply that the national system of innovation approach is without serious problems, particularly in relation to exploring the current changes in the global architecture of production. These are explored in the section below.

2.4.1

NIS and National Geography

The seemingly unquestioning acceptance of national borders within the NIS framework ignores the differences generated by the scale and structure of countries. In the United States, the national Government funds, by global standards, a very large defence and medical research12 program. Thus, the USA’s Federal Government clearly has had a very strong long-term impact on the dynamics of the US innovation system. At the same time, California by itself is the world’s 6th largest economy. California has benefited from federal research funds and from the opening up of new urban spaces during the twentieth century – in a movement of population from east to west (see Saxenian 1994). Today, California has substantial clusters in manufacturing and services with a major share of world demand for ICT equipment.13 It thus appears that California has benefited both from being included within the overarching political structure of a large nation state (itself the world’s largest economy) and from regional spatial agglomeration processes. The NIS approach seems to both suffer from, and to continue a confusion caused by the lack of long-term statistics at the sub-national level.14 The nation state is the dominant statistical feature of our understanding of global economics. International trade data is generally collected at the level of the nation state and not regions. Other data, such as industry value added or research expenditure for some countries, also becomes problematic below that of the national level. This availability of data biases analysis towards nation states. So whilst the emphasis of the NIS approach is on the uneven development paths of nations, even Freeman (2002, p. 209) points out that uneven development exists within countries. Therefore an interesting possibility it that the irregularities in development of long-term specialisation

12

AAAS (2004b). The latest estimates reveal that the USA has a significant share of the world market for ICT (European) Information Technology Observatory (EITO) 2004 estimates that the USA has 32 per cent of the world market and all of Europe (including the East) has 30 per cent of the world market. Analysis of Bureau of Economic Analysis data suggests that California represents approximately 17 per cent of the USA’s value added in electronic and electrical equipment. 14 Classifications of ‘regions’ still suffer from problems – see Casellas and Galley ‘The EU Nomenclature of Territorial Units for Statistics, referred to by the French acronym NUTS, is very heterogeneous in character. Tiny islands, cities, large rural regions and entire countries are considered to be comparable units for analysis’ (1999: 551). 13

30

2

Systemic Innovation and Nation States

patterns (knowledge accumulation and industry) seen in NIS studies may well be present at the level of regions (states or, provinces) but these issues are little analysed due to data deficiencies (see Chap. 3, for discussion).

2.4.2

NIS Politically Defined

Freeman’s view that the ‘phenomena of forging ahead, catch-up and falling behind in nineteenth and twentieth centuries can most plausibly be explained in terms of national systems’ needs to be tempered as such a generalisation ignores the changes to the power, borders and nature of those nations (2002, p. 209). As Elam writes: While technologies have been successfully portrayed as fluid and always in the making, nation-states have been largely accepted as fixed, stable and ready made. What has escaped attention is that just like technologies, nation-states are also being continually envisioned, designed, launched, remodelled, renamed, disassembled and scrapped. By failing to take adequate account of the historical contingency of modern nation states, research on national systems of innovation has been handicapped in its attempts to grasp contemporary phenomena such as globalisation and European integration (1997, p. 157).

Borders and constitutional political power evolve over time and may change more quickly and more often than those in the EITC field appear to acknowledge. The powers of the European Union have been in a constant state of flux since its birth in the post World War II period, as Table 6.8 makes clear. Each change and each enlargement of the European Union alters the dynamics of development and the process of change is unlikely to stop in the near future with ongoing negotiations for a written constitution and the enlargement process is envisioned to continue beyond the 10 countries that joined in 2004. The 10 new 2004 members plus the two 2007 new members of the EU will alter the dynamics of the pre-2004 economies in ways that are not entirely clear (see Gorzelak and Jalowiecki 2002). Recent research suggests that supra-state structures can play a role in promoting innovation and development15 through a re-distribution of funds to lagging regions and even assists in determining nation state borders.16 Although their impact is less, multilateral agreements on trade and intellectual property17 can influence the commercialisation of innovation and the catch-up processes. Only a few authors seriously suggest systems of innovation can extend beyond national borders. Even fewer studies can be identified that actually conduct supra-national research. The OECD identifies the concept of worldwide systems but notes that ‘national characteristics and frameworks always play a role in shaping 15

See for example the discussions in Cappelen, Fagerberg and Verspagen 1999 and 2000 on the effect of European Union structural funds for promoting regional development. 16 The Economist (2003c) When small is beautiful (p103) argues that the trend in recent examples of nation-state formation is towards smaller rather than larger borders. Crucially, however, it suggests that this might be due to supra-state structures such as the European Union that offer free trade zones across borders and other benefits of Federalism. 17 See for example Turpin (2000) for an interesting discussion of IPR in Asia Pacific countries.

2.4

Nations in a Changing World: Weaknesses of NIS

31

them’ (1999c, p. 23). However, despite the acknowledgement, the study does not actually identify any specific research that fits such a category. Elsewhere the term ‘supra-national systems of innovation’ has emerged. It is used by Bergman, Charles and den Hertog (2001, p. 9) and by Edquist (1997b and 2001). Few references are provided, however, it is typical to conceive of the supra-nation system as the entirety of a distinctive political system or jurisdiction. As Edquist indicates, ‘one may – in Europe – distinguish between a supranational system at the European Community level, the national level, and the regional/local level’ (1997b, p. 16). The examples of both Caracostas and Soete (1997) and Gregersen and Johnson (1997) both support the principle that innovation systems researchers have a tendency to think in terms of politically defined structures as the starting unit of analysis before focusing on the working dimensions of the ‘system’. The latter commenting that the ‘European system of innovation only exists, so far, in a rather narrow sense’ (1997, p. 489). Rather than focusing on potential cross-border functional sub-systems, they were testing for the rather more nebulous idea of generalised European integration. Such a perspective excludes the possibility that there might be supra-national systems that encompass, for example, the Scandinavian countries (noting that Norway is not a member of the EU) or that Germany and France might have close ties without the need for all the countries in the EU to have ties with each other in a European System. Such unity is unlikely to occur even within a single country’s borders. The political definition of systems also excludes analysis of cross-country economic links, which change structure depending upon the sector. Therefore, promoters of the NIS perspective over-emphasise the position of countries within the global economy. There are too few analyses of the causes and drivers of technological fragmentation (see Pavitt 2003a, b,18 and Chaps. 6–9) – where technological products are integrated from across the world. Instead policy recommendations such as those suggested by Archibugi and Iammarino (1999) are designed to strengthen individual countries (Table 2.1). Table 2.1 A response to globalisation: widening national technology portfolios Categories

Targets

Instruments

International exploitation of national innovations

Inward flows Achieving lower foreign dependency and filling technology gaps, increasing learning.

Incentives to infant industries. Promoting collaborations between national firms and leading firms in the field. Incentives to selected FDI in the country.

Source: Archibugi and Iammarino (1999, p. 327). This is a row from Table 7: Public policies targets and instruments for the globalisation of innovation.

18 Pavitt in a series of articles on globalisation discusses the shift of manufacturers in advanced economies from owning factories to focusing on design and product integration. Pavitt foresees a likely shift of manufacturing to developing economies but does not discuss the possibility that components or complementary products sourcing may move across (the) border, but be built in developed economies.

32

2

Systemic Innovation and Nation States

Given the evidence, already presented in this chapter regarding the difficulties of shifting technological or industrial specialisations, the advice here to ‘fill technological gaps’, is surprising. The technological complexity of products is increasing and thus the ability of countries to specialise in all the components for a given product would appear to be decreasing.

2.4.3

NIS, Borders and Economic Space

Carlsson puts it so simply ‘in view of the fact that most studies of innovation systems focus on national innovation systems, it is not surprising that little direct evidence is found that innovation systems are becoming global’ (2003, p. 20). Carlsson could uncover only a few analyses of the internationalisation at the spatial scale of systems. The analysis presented here is in agreement with Carlsson’s assessment of the literature. However, to simply expand the definition to larger but, nevertheless, politically defined territories such as the European Union also seems to be a dead end analytically. Many of the findings of the national innovations systems literature appear to be strongly substantiated by empirical evidence. What is strongly disagreed with here is the lack of attention that has been given to explaining the increasing prevalence of economic links and embodied knowledge flows (and innovation flows?) that are crossing national borders. The big question for those that promote NIS over other approaches is, if, as Edquist (1997b) notes, interdependency is central to the very essence of innovation theory, then why are interdependencies bounded by national borders? Niosi and Bellon (1994) is the only major exception that could be identified for this study. They identified a trend of increasing interconnectedness in the development of national systems of innovation as companies and scientists interact and move across borders. Too frequently, data results of cross border innovation patterns are presented as a general preference for national systems or proximity without the means to look for specific spatial structures in the data that might represent functional systems of innovation that extend internationally at important levels of scale. Although the nation state is an enduring politico-economic phenomenon, to construct the analysis of the massive changes in political and economic power that is occurring with the rise of East Asia and China, while existing economies remain prosperous, purely around nation state entities ignores critical dimensions of the interdependencies between countries. It is argued throughout this analysis that there are clear theoretical, empirical and methodological reasons to choose an analytical tool that allows the strengths of linkages between two places to define the spatial cores of interacting systems. Analysis should retain a spatial and therefore a systems perspective as is argued in the next chapter, which explores the research on sub-national systems (clusters and regions, etc.).

Chapter 3

Innovative Regions, Clusters and Milieux

‘Innovative networks cluster in different parts of each national economy, and seem related to the economic structure and their location seems to stay relatively stable over time. This is why the national – perhaps regional – systems of innovation matter’ (DeBresson et al. 1998, p. 4).

3.1

Innovation Territories

In Chap. 2, it was shown that some characteristics of nations have been important for determining the level of knowledge generation and technological innovation. The processes of knowledge accumulation, for example, are influential in the coevolution of the scientific and industrial specialisations of countries. Although, the level of innovative effort of countries is not pre-determined, it does seem that trajectories once established are hard to alter relative to that of other countries. An analytical perspective that adopts as its starting point the nation state, can be shown to provide valuable insights as to why particular pieces of geography succeed more than others at generating, acquiring and utilising knowledge. However, Chap. 2 also revealed that this national approach to the study of innovation has only partially succeeded in developing tools that can analyse the changes occurring in the geography of global production (see for example Friedman 2000). The emphasis on endogenous capability and the comparative benchmarking approach inherent in NIS analysis also highlights the need to build a multi-spatial innovation framework that can integrate the continuing benefits of place and the growing levels of trans-border production. This chapter takes the next step by focusing on the research which has investigated the competitiveness of specific places, whether they are economic agglomerations (clusters) or regions (political jurisdictions). This ‘agglomeration’ or ‘clustering’ is important to those with an interest in innovation because some locations are clearly more prosperous and successful at bringing innovations to the market. Why do firms in similar or related industries tend to locate themselves in close proximity to one another and what benefits are there from this practice? The research on this topic within the neo-Schumpeterian field can be broadly labelled as sub-national

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_3, © Springer-Verlag Berlin Heidelberg 2009

33

34

3

Innovative Regions, Clusters and Milieux

systems of innovation (Freeman 2002) and as clusters, industrial districts and regional competitiveness (Kitson et al. 2004) in economic geography. As with the analysis of the national innovation systems research, the discussion in this chapter on the importance of regions, clusters, sectors and industrial agglomeration is focussed on furthering an understanding of the connections between particular places and external locations. By firstly comprehending the dynamics of clusters and the value of proximity for the producers and users of knowledge and products, it is possible to draw conclusions on the significance of endogenous capabilities and the characteristics of extra-territorial (intra-national or international) linkages in knowledge and products. Given the breadth of the topic, Maskell (2001) is a particularly useful place to start. Maskell analyses the most frequently discussed theme in the sub-national systems innovation literature – clustering, observing that there is not yet a formulated theory of why certain industries group together in particular places. Further he argues, for there to be a theory in the future there must be progress towards ‘an explanation for the existence of the cluster’, ‘an explanation for the growth of the cluster’ and methodologies for defining ‘the boundaries of the cluster by specifying why the clustering of some economic activities precludes the integration of others’ (2001, p. 937). These three dimensions cover most of the important questions and provide a way of assessing the existing literature. In this chapter, the analysis of the research literature is constructed around the following themes (see also the literature framework diagram Fig. 3.1): • Locating clusters at different spatial scales (existence and definitions, etc); • The geography of agglomeration; • Cluster theory (the driver of proximity – traded and untraded interdependencies); and • Cluster boundaries and extra-territorial linkages.

Fig. 3.1 Cluster-based systems of innovation

3.2

Locating Clusters

35

Significantly, for this book, the framing of Maskell’s questions reveals the intraregional bias of the innovation discipline’s analysis of clustering on the basis of ‘differentiating between local ‘sticky’ knowledge and globally ‘ubiquitous’ knowledge (see Doloreux 2004). This is in contrast to the possibility of a framework that integrates local knowledge and inter-regionally differentiated knowledge that is spatially structured within networks.

3.2

Locating Clusters

As early as the mid-1990s, Storper identified a decade-long resurgence of interest in local development dynamics (1995), but in the decade since then a very large volume of research on the importance of location has emerged from economic geographers and increasingly from neo-Schumpeterians as well. It seems plausible to suggest that for some researchers interested in the processes which foster innovation, this new focus on regions was a reaction against the sole reliance on the national approach of NIS. For economic geographers, it was the re-discovery that, even with globalisation, place still matters. The main concerns of regional development, economic advantage, clustering and the role innovation and knowledge are common to both disciplines. ‘Sub-national systems of innovation’ (Freeman 2002) is now the topic of a burgeoning literature. One advantage of analysing the development of specific regions is that, while it can be the case that statistical analysis of national economies becomes too abstracted from the business realities, identifying territorial concentrations of industry, particularly of international scale, can give recognition to individual businesses as well as institutional and technological histories. Many places of industrial strength have names that have become brand labels for ‘their’ industry. A few from the past and present include: • • • • • • • •

Detroit – car manufacturing; Hollywood – motion pictures; London – banking and insurance; Manchester – cotton production in the first phase of the industrial revolution; New York – global financial services; Seattle – aerospace and software; Silicon Valley – information and communications technologies; and Wall St. – the USA’s stock market.

Such examples of global industry capitals1 provide iconic symbolism for the study of clusters, but can hide the fact that regional agglomerations2 are more 1

Klepper (2001) refers to Detroit as the capital of the US’s automotive industry. The idea seemed a particularly good one to borrow and apply on a broader basis. 2 The issue of differing scales of economic agglomeration is taken up in the last section of this chapter.

36

3

Innovative Regions, Clusters and Milieux

ubiquitous than unique. Porter (1998 and 2003), for example, provides an analysis of clusters he has identified across the USA using a location quotient methodology. His results highlight concentrations of particular sub-branches of industries, such as carpet making or instrument making, which can reside in small regional cities. All of these industries and locations have unique histories of development that led to the agglomerations visible today.

3.2.1

Clusters, Regions and Sectors

It is important to recognise, that while there is a growing body of research on the economics of innovation in specific places, there are still debates as to the appropriate geographic and industrial level of analysis. As Cooke notes, some analysts strongly advocate a national system approach. ‘Because innovation systems analysts have been wedded to sentiments of ‘national’ economies, the concept of a RIS [regional innovation system] is a relatively new one … The development path of the concept was almost entirely from regional science and economic geography. Indeed, authors like Lundvall (1992) were strangely hostile to the concept’ (2001, p. 949).

Cooke is particularly attracted to politico-administrative regions (see especially Cooke 2001 and Morgan 1997) as the appropriate scale of analysis but industrial agglomerations (clusters), sectors, industrial districts and other labels and relevant foci also exist in the literature. Clustering can refer to groups of firms producing similar products3 at a similar stage of production stage or can refer to firms spread across horizontal or vertical supply networks.4 Clustering can also refer to situations where knowledge infrastructure (universities and public research laboratories) are co-located with relevant branches of industry. Alternatively, sectoral innovation systems (Malerba 2002) and product-service systems (Marceau et al. 2001), which are similar frameworks; focus on analysing a range of characteristics, including government policy, knowledge infrastructure, training institutions and industrial activity, R&D and suppliers and users. Malerba’s ‘sectoral systems of innovation and production’ concept, for example, is defined as: ‘a set of products and the set of agents carrying out market and non-market interactions for the creation, production and sale of those products. A sectoral system has a specific knowledge base, technologies, inputs and demand’ (2002, p. 247).

At the present time, sectoral systems as a concept has not caught on, compared to the burgeoning analysis of clusters. Although the concept could be aspatial it does seem to be applied sub-nationally. The concept of technological systems

3 This can be identified by showing which locations have a higher concentration of industry activity (through employment statistics or value added) when compared to other places. 4 This has often been analysed with the use of input-output data that facilitates the identification of industries which are related through supply chains within a particular location.

3.2

Locating Clusters

37

‘in which the focus is mainly on networks of agents for the generation, diffusion and utilization of technologies’ (Malerba 2002, p. 248) is not considered here. It has a focus on the technical skills and communities associated with specific technologies, rather than industries, although analytically it has spatial elements. According to the OECD (1999c) traditional neo-classical economic analysis over emphasises price-based competition, products for final consumption (rather than intermediate goods and services), and businesses are primarily viewed as entities largely independent of other firms and organisations such as universities or government research laboratories. By contrast, research on sub-national systems of innovation emphasises government policy, publicly funded research, and interdependencies between businesses and between businesses and universities in the co-location of corporate development. In one way or another, all of the factors listed by the OECD under clusters analysis have a focus on the interdependency of the different actors. The OECD is not alone in noting this distinction of ‘competition over cooperation’ (see in particular De la Mothe and Paquet 1996, p. 11). As was noted in Chap. 2, interdependency is an important factor of the NIS framework. At the scale of regions and clusters, a range of different kinds of interdependencies are no less important but probably easier to measure.

3.2.2

Defining and Finding the Places of Interest

As might be expected, from the range of labels for the phenomenon of innovative agglomeration just listed, there is no agreed methodology for identifying where such clustering occurs or what constitutes these districts. The definitions and the language of description for clusters are still badly under-developed. Clusters have been portrayed as ‘high-productivity, knowledge-rich, decentralised, entrepreneurial, and socially progressive economy within the reach of local policy makers’ – in effect as Martin and Sunley (2003, p. 29) note, clusters are seen as a policy panacea. Other authors have expressed similar sentiments but Martin and Sunley most eloquently state ‘the cluster literature is a patchy constellation of ideas’ some ‘clearly important’ and ‘some which are either banal or misleading’ (2003, p. 28). In effect clustering is a ‘brand’ label so ‘elastic’ (p. 29) to the point of being almost universally applicable. More critically, Martin and Sunley observe that an association between high growth industries and geographic concentration does not mean the concentration is the cause of the success and whilst the vision of local economically prosperous industries is attractive; the evidence is still in its ‘infancy’ (2003, p. 29). Regional systems are normally more easily defined because, typically, they coincide with politico-administrative spaces such as sub-national states and provinces (or occasionally urban areas). The term ‘cluster’, however, has been attached to every scale of economic agglomeration – everything from a few businesses or scientific labs in a small town through to the national ‘mega-clusters’ (see Roelandt and den Hertog 1999, p. 14). All such ‘clusters’ are identified with a wide variety of methodologies (see for example Roelandt and den Hertog (1999, p. 16). It is thus

38

3

Innovative Regions, Clusters and Milieux

important to have a way of quickly categorising existing research and clearly identifying the approach used in the research completed for this thesis. Only a few authors have attempted to bring some order to the classification of cluster types. According to Verbeek (1999), there is one broad division in the clusters literature. The first class of clusters is based on similarity, which includes the analysis of geographic concentrations of industries. The second class focuses on interdependence, chains of production and filières. This is not a completely satisfactory typology as it neglects to focus on geographic scales of activity. Unfortunately, both Drejer et al. (1999, p. 294) and Roelandt (1998, p. 8) who attempt to give greater weight to spatial scale and to adhere to some degree to the position proposed by Verbeek, have confused different types of data and different geographic scales (see Table 3.1). At the meso and national levels, this presentation creates a number of confusions. At the national level, it suggests that the cluster concept includes linkages but focuses on specialisation patterns. As data on industrial interactions and specialisation are quite different, it is unclear how this schema works, or how it would be implemented. At the ‘meso’ level, the authors do not explicitly identify a geographic scale for the spatial distribution of the branch of industry and could therefore effectively be a national spatial range. Finally, it is worth noting that the content of the table implies the use of both statistical data and qualitative data across both columns. This attempt at clarifying the analytical possibilities does, it seem, only add to the confusion surrounding the identification of clusters. Such vagueness between the different cluster concepts, data types and geo-political scales would suggest it is necessary to attempt to clarify some of the language. In particular, some transparency in distinguishing between the two dimensions of industry and geography would seem to be helpful. Table 3.2 presents an attempt at just such break down which are then used throughout the rest of this thesis (as much as is reasonable). Table 3.1 Cluster analysis at different levels of analysis Level

Cluster concept

Focus of analysis

National level (Macro)

Industry groups linkages in the economic structure.

Specialisation patterns of a national/ regional economy. Need for innovation and upgrading products and processes in mega clusters. SWOT and benchmark analysis of industries.

Branch or Inter and intra-industry linkages in industry the different stages of the level (Meso) production chain of similar end products. Firm level (Micro)

Specialised suppliers around one or a few core enterprises (inter-firm linkages).

Exploring innovation needs. Strategic business development.

Chain analysis and chain management. Development of collaborative innovation projects. Source: Roelandt (1998: 8).

3.3 The Geography of Agglomeration

39

Table 3.2 Cluster typology: industry statistics and geographic nomenclature Analytical approach including data

Geographic descriptor

Specialisation: Value added, Nation state production, export or employment data. [National industries such as the auto industry etc]. Specialisation: Value added, production, Regions (Provinces/States) export or employment data. [Location quotient methods]. Specialisation: Value added, production, Cities, urban complexes or export or employment data. [Location other specific sub-regional quotient methods]. locations.

Name National-macro clusters

Regional-macro clusters Urban-macro clusters

Inter-industry and intra-industry linkages. Input-output analysis, supply chains etc. Inter-industry and intra-industry linkages. Input-output analysis / supply chains etc.

Nation state [national I-O or similar data]

National-meso clusters

Regions (Provinces/States) [regional I-O or similar data]

Regional-meso clusters

Business to business networks Qualitative evidence of business connectedness. Business to business networks Evidence of business connectedness and supply chains in specific locations.

Nation state (business networks National business with no particular geographic networks locus) Cities, urban complexes or Industrial district other specific sub-regional networks locations.

A term such as ‘macro’ might be used to refer to clusters defined by relative specialization patterns. Such analysis might use location quotients to compare the strength of one industry against others across geographic locations or apply other techniques. For clusters based on interdependencies, terms such as meso or ‘web’ might be appropriate. The definable geographic scales might be nations, regions (states, provinces) and urban-regions, but these are only examples. Therefore as the empirical research for the present thesis has been based on inter-country (national) input-output (inter-industry linkages) data, the individual clusters have been defined as ‘national meso-clusters’. This categorisation has problems of its own but it makes progress towards a necessary clarification of spatial and data types, if we are going to begin to develop meta studies of clustering.

3.3

The Geography of Agglomeration

Comparing how (sub-national) regional strengths evolve across time has been, very difficult until quite recently, in comparison to analysing national development trends due to data limitations. There is a growing opportunity for improved analysis

40

3

Innovative Regions, Clusters and Milieux

of regional activities, advantages and trends as national statistical agencies focus more of their effort on collecting data at various spatial scales. Across the European Union, there is multi-country structural pattern of uneven regional development. Combes and Overman (2003) identify a strong pattern of core and periphery5 regions. When analysed for their proximity to markets (GDP per capita), regions in Western Germany, Northern France and South East England are revealed as the concentrated zone of value added (GDP) for Europe. The authors reveal that beyond this core there is a graduated drop in GDP per capita as distance increases. In these EU economies, regional income converged between countries in the period 1980 to 1999, but regional income disparities within countries widened (see Le Gallo and Dall’erba 2003). Interestingly, the degree of industry concentration in the USA (at state level) and Europe’s regions6 is not significantly different, but in Europe, the trend has been towards greater regional concentration more rapidly than in the USA (see Andaluz et al. 2002). The underlying drivers of such trends are still hotly debated. There are many arguments for the economic significance of regional concentrations but one of the more important ones is that knowledge generation is spatially agglomerated and knowledge diffuses poorly across distance (see the next section on theories of proximity). The cumulative nature of knowledge (explored in Chap. 2) and the link between the generation of knowledge and the ability to benefit economically is likely to be one set of characteristics that contribute to uneven development.

3.3.1

Knowledge and Innovation Geography

GDP, not the only characteristic of economies that is highly agglomerated, patenting activity is as well. Eurostat (2002) has analysed the geographic dispersion of patent applications for the EU 15 group of countries across statistical regions. Table 3.3 lists the top 15 regions for high technology7 patent applications (per mill of labour force). In 2000, there were nearly 57,500 applications to the European Patent Office (EPO) for patents from all regions within the EU’s (then) 15 member states. This included 10,500 applications for patents in high-tech industries. The geographic distribution of the patents is highly concentrated – ‘21 regions out of 211 accounted for more than half of all the patent applications filed with the EPO, and only 13 regions produced more than half of the high-tech applications’ (Eurostat 2002). In a

5 The analysis of inter-country input-output modelling presented in chapters seven, eight and nine of the present thesis provides further support to the core-periphery ideas, revealing that Germany is the economy in Europe most relied upon to supply industrial components across many industries. 6 75 statistical territorial units (NUTS) used included the country of Denmark and provinces in Germany, France, Italy, Portugal and Spain. 7 Aviation, computers and automated business equipment, communication technology, lasers, micro-organism and genetic engineering, semi-conductors.

3.3 The Geography of Agglomeration

41

Table 3.3 EU Regions with the most high-tech patent applications per-mill labour force (2000)

Member State NUTS 2 region

High-tech patent applications per million labour force

High-tech patent applications

High-tech patents as % of all patent applications

1 2

Germany Finland

540.9 530.4

1132 416

37% 52%

3 4 5 6 7

Netherlands Sweden Sweden Finland United Kingdom Finland Germany United Kingdom

Oberbayern Uusimaa (Suuralue) Noord-Brabant Stockholm Sydsverige Pohjois-Suomi East Anglia

524.2 430.0 336.3 312.1 236.3

633 416 199 86 265

40% 40% 35% 54% 39%

Etelä-Suomi Mittelfranken Gloucestershire, Wiltshire & North Somerset Hampshire & Isle of Wight Stuttgart Övre Norrland Oberpfalz Ile de France

202.4 189.7 169.6

188 160 197

37% 19% 39%

169.0

156

43%

162.9 160.5 159.6 155.1

315 39 84 854

12% 35% 20% 25%

8 9 10

11

United Kingdom

12 13 14 15

Germany Sweden Germany France

Source: Eurostat 2002.

similar vein, Paci and Usai (2000) analysed innovation (patents per million inhabitants), labour productivity and industrial specialisation in 109 regions in Europe reporting that: ‘there is a tendency towards the formation in Europe of highly specialized technological enclaves, especially in some sectors – machinery, transport equipment and energy. Moreover, we have documented how the spatial and sectoral specialization of innovative and productive activities is positively and significantly correlated’ (p. 108).

Therefore, there is a link between the degree of specialisation in knowledge generation, seen in patent applications, and the pattern of industry specialisation. However, patenting may not provide a good indicator of innovative behaviour because a few urban centres may be the base for the major corporate research and development facilities and thus be the sites for patent initiation. This would not indicate whether, innovation is geographically agglomerated. By analysing data from the European Community Innovation Survey on businesses that identify themselves as innovating and identifying them by region, Evangelista et al. (2002) reveal that there were only a limited number of areas of Italy which could be described as ‘regional innovation systems’. Beaudry and Breschi add an important dimension to this finding. They report that firms are more likely to be innovative if they are

42

3

Innovative Regions, Clusters and Milieux

co-located with other innovative firms and in the ‘presence of spillovers associated with a large accumulated stock of knowledge’ (2003, p. 339). The converse is also true – disadvantages arise from the ‘strong presence’ of non-innovative firms in a cluster. Beaudry and Breschi found for Italy, but not for the UK, that the presence of firms in related industries enhanced innovativeness. Thus, there are economies to innovativeness, in that, regions that are innovative, stimulate more innovative activity. Finally, concentrations of entrepreneurial innovativeness revealed through new firm births had better employment growth prospects than places that have a higher reliance on large corporations in a comparison of 74 districts across West Germany (see Audretsch and Fritsch 2002). Clearly, there is strong evidence that production and innovation clusters within particular locations in economies are related. There is also good evidence that there are strong dynamics that drive a tendency for knowledge to be localised. These findings are not in dispute here, what matters in this book, is not that clustering occurs, but the importance (the role, scale and spatial structure) of external links which, it is argued here have not been thoroughly explored, to date. A general tendency towards proximity for knowledge and industry is the not the same, as specifying how those agglomerations fit within multi-spatial innovations systems, e.g. the systems of systems.

3.3.2

Supply Architecture

As innovation and production clustering appear related, data on value chain interdependencies can provide useful descriptive information. Measuring production interdependencies at both the national and regional levels has been used to locate clusters (economically and geographically) and to describe the scope and density of relations. Input-output (I-O) methodologies have been used widely (see for example OECD 1999d and 2001c) to map value chains because it can trace the flow of intermediate goods and services. Such flows are the supply and use of industrial ingredients which are required for further processing. At this point, to analyse value chains through a time dimension with I-O data is most easily conducted at the national scale. Most regional analyses exist only as one off studies. At the regional level, Trends Business Research (2001) utilised I-O and value added data to identify clusters in the United Kingdom. Feser and Bergman (2000) used national input-output data and economic data for the American State of North Carolina to create information on regional supply chain clustering patterns in that State. The benefit of their approach is that it identifies inter-industry linkages and thus vertical and horizontal linkages at the sub-national level rather than just relying upon location quotient approaches. In research, that could prove to be complimentary to existing approaches that map industry inter-linkages, Hoen (2002a) has developed a methodology for calculating the importance of particular linkages to potential clusters and is thus able to determine, statistically, linkages that are within a particular cluster and those outside. Not only can I-O data be

3.3 The Geography of Agglomeration

43

used to understand the web of industry relationships (see Haukness 1999) at a point in time, it can also be used to measure changes across time. Wixted (2005) has established that at the national level, business to business relations (intermediate input-output relations) were relatively stable during a 20 year period (1970–1990) for 9 countries. The charts reveal important changes, such as the growth of value chains based around service industries generally, and business services in particular. With a view to the question behind the present research, the charts provide some evidence that tempers the commonly held view that western economies lost significant amounts of manufacturing activity during this period. Supplies to manufacturing industries, often from other manufacturing industries (which are obviously not the same as goods for final consumption), often grew at the same rate as the economy overall and thus remained surprisingly stable as a share of GDP. A number of general characteristics appear from Wixted’s analysis of OECD economies with matrices of 33 industries by 33 industries. These include: • Mostly, intra-industry connections for manufacturing could be seen as generally stable with the rate of growth of GDP, with declines of typically (outside of some in Japan) at less than one per cent of GDP for the period; • The largest apparent decline was for intra-industry iron and steel supplies in Japan; • There was a widespread decline in textiles and clothing industry supplies, mostly to the textiles and clothing industry; and • Supplies growth in the service sector is clearly visible for most countries – particularly in industries such as business services. Input-output relations, the structure of ingredients that it takes to make a given product, look to be much more resilient to dramatic change when compared to other variables such as employment or an industry’s overall share of GDP. Total manufacturing employment declined in many OECD countries during the 1970s and 1980s, (see Godbout 1993). Manufacturing’s share of GDP for OECD countries8 has also been constantly declining, shifting from 29.1% in 1960 to 28% in 1968 to 21.6% in 1990 and then to 19.9% in 1995. Thus, within the space of 20 years, manufacturing’s share of GDP had declined by 22.9% for the OECD (a share that is representative of the decline in individual countries – see OECD 1999e: 67). The comparatively stable nature of inter-industry linkages is supported by Verspagen who notes that for the US economy, the linkage structure is rather sticky. ‘Rank correlations between forward and backward linkages of sectors over periods of roughly 10–15 years are rather high. This also explains why industries related to ‘old’ technological revolutions dominate the linkage structure for a long time’ (2002, p. 11).

8 Canada is revealed as the only exception to declines in the absolute levels of employment in manufacturing in a study covering the USA, Canada, Australia, Japan, France Germany, Italy, the Netherlands, Sweden and the UK.

44

3

Innovative Regions, Clusters and Milieux

It should be noted that the import intensity of manufacturing production is generally increasing, indicating that more components are necessary to make the same products. It can also be observed that within the literature on innovation in regions and clusters, where input-output data was used, there were no examples of it being developed within a multi-spatial (regions or countries) framework. A few examples of analysis of inter-city trade linkages have been described. These are discussed in Sect. 3.5.2.

3.4

Cluster Based Theories of the Drivers of Proximity

The recognition of the tendency for economic activity to agglomerate in particular places has been written on for over a century. The intellectual history is often traced back to Alfred Marshall’s discussion of industrial districts in the 1890s. His commentary on industrial districts was particularly focussed on the significance of the advantages accruing from large pools of people with specialised skills that can exist in local labour markets due to many businesses in a similar industry co-locating. It seems, however, that for most of the 20th Century there was little interest in the economics of territories (economic places whether regions or other definitions). Baumol (2000) argues that the big changes in the discipline of economics after Marshall were in the formalisation of macroeconomics and the contribution of Schumpeter to initiating the interest in technological innovation. It is worth noting that Baumol shows no interest in the industrial location contributions of Marshall, perhaps emphasising the current economics textbook focus of macro, micro and trade theory. Weintraub (1999) would seem to agree. In a review of the important revolutions of twentieth century economics thinking, he suggests that many important changes to economic ideas emerged during the century. The changes are catalogued under various headings, but neither innovation nor regional development is included in the lists. In contrast, Craft’s review (1999) of twentieth century economic growth research, reveals much more interest in the role of R&D, technology and innovation as a driver of economic growth, although the focus is on national growth and not territorial development. These reviews shed some light on the lack of interest in innovation and place. For economic theory, the excitement has been largely elsewhere – welfare economics, Keynesian macroeconomics and the monetarist response. One problem for neoclassical economics is that the Heckscher-Ohlin theory of trade is not a theory of location as Venables points out: ‘The spatial unevenness of industrial development requires an explanation beyond that offered by comparative advantage … Heckscher-Ohlin theory is, unsurprisingly, not good at explaining the location of industry across areas where factor endowments are broadly similar (as in much of Western Europe)’ (1998, p. 2).

In contrast to this deficiency in traditional analysis, a growing number of researchers are keen to pursue Marshall’s interest in the economics of place. Storper attributes the re-emerging interest in regional economics to a growing

3.4

Cluster Based Theories of the Drivers of Proximity

45

recognition that the differences at the local level might be both important and informative. ‘the region might be a fundamental basis of economic and social life ‘after mass production’. That is, since new successful forms of production – different from the canonical mass production systems of the post–war period – were emerging in some regions and not others, and since they seemed to involve both localisation and regional differences and specificities (institutional, technological), it followed that there might be something fundamental that linked late twentieth-century capitalism, regionalism and regionalization’ (1995, p. 191).

Hall in his monumental analysis of the development of cities on various continents in various ages has the following to say about Detroit during the early years of its growth: ‘The question still must be: why did Ford arise where and when he did? He could have been born anywhere, and maybe could have succeeded anywhere; there was no inevitability that put his name and Detroit’s together. Yet there was a strong probability. Detroit in the 1880s and 1890s was at the industrial frontier. It was not one of the great industrial cities of America, like New York, Philadelphia, Pittsburgh or Chicago. But neither was it at the edge of the world. It was growing very rapidly on the basis of natural resources around it and of the industries that had sprung up to transform those resources into products. It was thus a rich and varied engineering centre, and it specialized, above all, in producing transportation equipment; it was one of America’s major railroad car manufacturers. The big firms subcontracted jobs to smaller ones, so there was a rich network of interdependencies and skills’ (1998, p. 423).

According to Hall, Detroit gave Henry Ford three advantages; venture capital, ‘fellow spirits’ interested in the potential of the motorcar and a network of contractors (1998, p. 423). These three features of Detroit in the 1890s are the characteristics which are commonly attributed to development of clusters. The following sections cover the role of the ‘fellow spirits’ in passing on information (untraded interdependencies) and traded interdependencies for the flow of components are discussed in depth. The section also analyses the role of business supply relationships (user-producer relations) for their role in promoting innovation. Many of the characteristics of innovation and knowledge considered important to the operation of national innovation systems, outlined in Chap. 2 can be applied to regional levels of analysis, the basis for an emphasis on regions and clusters is that firms gain particular benefits from proximity. Amongst others, Cooke and Morgan (see Cooke and Morgan 1998, Cooke 2001, Cooke et al. 2004, Morgan 1997, 2001) have particularly emphasised the role of the milieu of local businesses, universities, policy and institutional arrangements for promoting innovative and competitive businesses.

3.4.1

Traded Interdependencies: Supply Architectures

One reason why companies might co-locate parts of their production system with those of other businesses in the same industry is to benefit from the synergies of cheaper access to components, quicker delivery times and perhaps cost effective

46

3

Innovative Regions, Clusters and Milieux

access to raw materials arising from a sufficiently large demand base. However, supply chain interdependencies are not often seen as a major factor of co-location outside of a few industries such as the passenger motor vehicle manufacturing (see Riemens 2002 and Gereffi 1998). More frequently, the innovation literature points to other factors such as untraded interdependencies in knowledge and access to knowledge which provide the incentives for industrial agglomeration. The structure of supply chain networks are, nevertheless, the focus of a few authors and an important element in the descriptions of many clusters. In Porter’s view,9 (1990) there are three important regional conditions which together with appropriate corporate strategies can give rise to successful firms. The mix of demanding clients, access to industrial supplies and a deep market of skilled labour within regions, it was argued, provide the critical conditions for industry competitiveness. Porter presented a ‘diamond’ of factors that drove clustering (Fig. 3.2). Porter’s argument is that these factors work best within close geographic proximity, even though at times ‘clustering’ is considered in a national context. Porter’s analysis has, however, not been uncontentious. Yetton et al. (1992), for example, argue that the evidence does not support the argument and especially highlight the economies of Australia, Canada and New Zealand as three resource rich highincome countries that do not fit the manufacturing industry orientation of Porter. Rugman and D’Cruz 1993) are also interested in the economies of Canada and New Zealand, criticising Porter’s analysis for its lack of understanding when it comes to Firm strategy, structure & rivalry A group of domestic rivals encourages the formation of more specialised suppliers as well as related industries Demand conditions

Factor conditions

Specialised factor pools are transferable to related and supporting industries.

Large or growing home demand stimulates the growth and deepening of supplier industries Related & supporting industries

Fig. 3.2 Porter’s diamond Source: redrawn from Porter (1990)

9 Although there is much made of Porter’s ‘cluster’ analysis one of his main contributions to economic geography debates is his business strategy perspective. The regional science field has long been researching the agglomeration of industry clusters (see Czamanski 1971 and Czamanski and Czamanski 1977) and complexes (Isard 1959). This regional science contribution is not recognized by Porter.

3.4

Cluster Based Theories of the Drivers of Proximity

47

the market operations of smaller economies. They suggest that rather than taking a nationalist approach, the functional market area would provide a better definition of the diamond. Thus, Canada and the USA would be part of a North American ‘double diamond’ (more on this in Sect. 4.2.3). In contrast to his earlier work, later research by Porter (1998 and 2003) has been increasingly quantitative, re-dressing the lack of data in the 1990 book and aims to show that clustering is not a special phenomena but a general economic feature of countries. The later work relies upon location quotient analysis which is an effective methodology for comparing the relative regional specialisation (urban-macro clustering) patterns based on statistical geographic units. Location quotients compare indicators (e.g. employment) for one area and compare them against the average for a wider geographic area (sometimes the nation). Porter’s work, increasingly, is more about the existence of clusters and their economic role than advancing the discussion as to why they exist. Only Steinle and Schiele (2002) present an analysis of the possible dynamics that generate the incentives for industrial agglomeration on the basis of supply industry interactions. They suggest that there are both necessary (NC) and sufficient (SC) conditions for clustering. • NC1: divisibility of process (including scale) – multiple specialist businesses; • NC2: ease of transport of final product & difficulties with transportability of components; • SC1: longer value chains • SC2: multiple dissimilar but complementary competencies • SC3: importance of innovation. Two of the five stimuli towards clustering appear to be related as the divisibility of the process and dissimilar competencies seem to be different expressions of the same concept. These conditions for clustering would suggest that the more complex the final product, the greater the incentive for firms to geographically cluster. Dissimilar but complementary competencies would often lead to the development of specialist businesses (possibly through outsourcing). According to Steinle and Schiele, longer supply chains for components and difficulties in transporting components in a particular industry would add to the conditions promoting co-location. Final goods, then with bulky components, in this structure would tend to be clustered. A notable exception might be the Airbus A380 being built in multiple locations across Europe and assembled in France (see Chap. 8). Other authors have put forward analysis that suggests that industries based on new knowledge (see Audretsch and Feldman)10 tend to cluster. Results presented 10 ‘Indeed, we find that a key determinant of the extent to which the location of production is geographically concentrated is the relative importance of new economic knowledge in the industry. Even after controlling for the concentration of production we find evidence that industries in which knowledge spillovers are more prevalent – that is where industry R&D, university research and skilled labor are the most important – have a greater propensity for innovative activity to cluster than industries where knowledge externalities are less important’ (1996: 639).

48

3

Innovative Regions, Clusters and Milieux

later in this book (Chaps. 6 and 9) would indicate that this theoretical construct needs to be extended to include inter-regional supply chains as entire clusters appear to be specialising in complementary products (Bresnahan et al. 2001). It is also worth noting that industries which broadly rely upon production scale for competitiveness (resource processing and transport equipment industries) account for the majority of highly internationalised industries reported here and analysed in Chap. 6). These results suggest that industries where it is cost effective to transport components but where there is either resource or skill complementarities, supply structures might elongate globally through networks. If supply chains are not typically thought to be the drivers of proximity, then what is?

3.4.2

Users and Producers: Linking Production and Innovation

The nexus between innovation and production (and later inturn to locality) rests with a fundamental characteristic of innovation. Innovation not often arises with the heroic sole inventor, but is rather due to the interactions between product and service producers and product users (DeBresson 1999). The finding that innovations, particularly those that require technological change, involve a relationship between businesses as users and businesses as developers is both robust and one of the few properties of innovation that can be related to data on economic structures. DeBresson et al. (1998, p. 39) emphasise that collaboration and cooperation are a prerequisite for innovation. Long-lived demand-supply contexts which build trust between the partners and tacit knowledge on the requirements of users and the capabilities of producers are significant for minimising the risks associated with innovating (see Lundvall 1992 and Fagerberg 1992 amongst others). In his study of a series of technological innovations, von Hippel (1988) emphasised that the creation of innovations was not a ‘market’ exercise but were often designed within a network. The inside information on problems and capabilities is what enables organisations and people involved to know enough to redesign, re-organise, or invest in new products. On occasions, organisations need to co-develop technologies. DeBresson has analysed the importance of the networks across a wide variety of industries (with Murray 1984) and more recently, he has argued that almost all innovations result from the linkages between users and suppliers (1999). His survey of innovative businesses has revealed that Canadian firms have as many as seven key users of innovations. Table 3.3 presents DeBresson’s results for just the ‘world first innovations’. In this category of innovative businesses, the highest percentage of firms (23%) co-innovate with three firms. DeBresson’s analysis of businesses either adapting innovations or developing ‘world first’ innovations reveals that there were no cases where other firms were not involved in the innovative activity. These findings are supported by the work of Porter’s qualitative (1990) and Fagerberg’s quantitative (1998) studies, that both argued that the participation of

3.4

Cluster Based Theories of the Drivers of Proximity

Table 3.3 Number of independent organizations required for each innovative act in Canada, 1945–1979 (DeBresson 1999)

49 No. of firms in innovation

Total % World first & adoption

1 2 3 4 5 6 7 Total

9 16 23 21 19 7 5 100 %

Source: DeBresson (1999, p. 4) used with permission

advanced domestic users of products was correlated to higher levels of industry competitiveness. Fagerberg makes a case that there was an association between the home market and industrial competitiveness for the period of (available data) 1965 to 1987. Drejer’s analysis of the hypothesis that the home market matters, suggests that there is a connection between input-output linkages and export specialisation. ‘Hence, it seems fair to conclude that both interindustrial linkages and technological activities in the nationally located industry are important in the determination of national export specialization patterns. However, the importance differs according to the mode of innovation in the industries, distinguished according to Pavitt sector characteristics’ (1999, p. 164).

This relationship-based paradigm of innovation, is in stark contrast to the market based approach of neo-classical economics. Instead of the destiny of businesses being dependent on their individual ability, DeBresson observes that networking combines capabilities and market positions – which are critical for the development of new products and services. ‘As no one organization can possibly keep internally all these dissimilar competencies, but tends to focus on similar competencies, innovation can only be undertaken through the collaboration of different enterprises. Yet the social sciences at the end of this century, except for a tiny part of economic sociology (for instance Burt, 1982, 1992), still only use the individual organization as the sole unit of coordination … This thesis also claims that innovative networks are now central to the understanding of all contemporary economic processes – not just innovation’ (1999, p. 2).

The finding that user-producer relationships are critical for innovation provides an analytical and theoretical bridge between the innovation at the corporate scale and innovativeness at the economy wide scale of activity as well as between innovativeness and the productive structure of economies. The latter is observable through, for example, input-output data, as DeBresson (1996) and DeBresson et al. (1998) have shown. Economic input-output structures can be closely (although not perfectly) related to the data available on the producers and users of innovations. Apart from the latter analysis, which makes some brief observations on the subject, there appears to be scant consideration of innovation based user-producer relationships extending across national borders.

50

3

3.4.3

Innovative Regions, Clusters and Milieux

Untraded Interdependencies: Knowledge Flows

Traded goods are based in contractual transactions, while user-producer relations develop through repeated use of those same channels for accessing components and are thus less formal. The most informal of influences on the existence of clusters is the flow of information one person to another. Although van den Berg et al. emphasise that clustering is a combination of production and knowledge institutions; it is the dynamics of uncodified knowledge which has attracted the greatest attention. ‘most definitions share the notion of clusters as localised networks of specialised organisations, whose production processes are closely linked through the exchange of goods, services and/or knowledge. In particular, the informal exchange of information, knowledge and creative ideas is considered an important characteristic of such networks. This is often referred to as untraded interdependencies’ (2001, p. 187).

As has been shown in Chap. 2, at the level of nation states the study of knowledge focuses on specialisations in science fields, as well as the path dependent trajectories of knowledge accumulation and national innovativeness. At the local level, the interest in knowledge is particularly engaged with the role of tacit knowledge in fostering a general climate of innovation. Tacit, as opposed to coded knowledge (information contained in scientific journals and books, etc.) is a catchall definition for different kinds of knowledge that cannot be repeatedly and reliably accessed. A number of authors have attempted to develop taxonomies of knowledge types. Johnson (1998, p. 10) brings many of these together for comparison. Johnson’s summary reveals that a number of authors attach key significance to tacit knowledge in various forms. Social knowledge, market knowledge, encultured knowledge are all important but can never be found written down completely and many activities need to be learnt by doing. Tacit knowledge is often focussed on how to do, how to fix, or who to go to (suppliers or markets). It is easy to overestimate how much production knowledge is recorded (outsiders expect it to be ‘written down somewhere’) in a way that could be replicated. Information is however, often specific to individuals as it is created from their individual experiences. Tacit knowledge is also often seen as problem-solving knowledge, as opposed to codified knowledge which is accessible to all but does not reveal all. Tacit knowledge is especially useful for obtaining techniques to get around technical barriers. Typically, ‘knowledge remains tacit if it is complex or variable in quality’ according to Lundvall and Borrás (1997, p. 14). It is often context rich and highly geared to particular situations. As such, it is locally relevant and increasingly valueless the further it moves from that specific context. Tacit knowledge is bound to humans who live and work in specific places rather than the geo-political spaces of nations or the global economy and it is more embedded within specific spatial/technological contexts. It is not just that tacit knowledge is in the heads of locals, it does not flow easily, quickly or far, passing often by word of mouth. It is therefore considered spatially limited. Saxenian attributed

3.4

Cluster Based Theories of the Drivers of Proximity

51

some of the early success of Silicon Valley’s businesses to local pub talk11 (1994, p. 32). She found that information on ‘competitors, customers, markets, and technologies’ was accessible, so ‘entrepreneurs came to see social relationships and even gossip as a crucial aspect of their businesses’ (1994, p. 33). However, such ideal types of clusters where there is a free flow of valuable information are probably comparatively rare. ‘Local knowledge spillovers’ might be mediated by ‘economic (market and non-market) mechanisms’ according to (Breschi and Lissoni 2001b: 270). In their analysis, instead of knowledge flows being gossip, swapped over a beer, the communications between engineers are informally regulated and traded. Breschi and Lissoni are, however, broadly in agreement with other literature on tacit knowledge, commenting that tacit knowledge has more effect on speeding up the innovation cycle than providing ‘new ideas’ (2001a: 1000). Despite the strong argument for learning (Malmberg and Maskell 2001) or untraded interdependencies as a key influence on clustering, it is unfortunate that the majority of the data that is used to assess clustering cannot measure knowledge flows. Holmen and Jacobsson do not criticise the assumption that knowledge spillovers are important, just ‘that when the theoretical basis for cluster formation largely lies on knowledge externalities, the methods employed to delineate clusters are based not on classes of knowledge but on product or industry classes instead’ (1999, p. 335). Breschi and Lissoni (2001a, b) agree and argue that an important research agenda for the future will be to follow the actual knowledge networks of individual companies and then map these against physical geography. One way this has been analysed already has been to use the citations of relevant scientific research included in patents to evaluate whether there are any regional patterns. Hicks et al. found that inventors have an inclination to acknowledge university research conducted in the same region (U.S. State). They comment that the bias towards authors of research in the same state provides evidence ‘that technological development has strong links to local scientific research’ (2001, p. 692). In a similar study, Verspagen and Schoenmakers tested the citing of patents within other patents to analyse intra-firm and inter-firm knowledge spillovers within a geographic setting. Their results ‘generally confirm’ that proximity matters for technological spillovers, finding ‘significantly negative coefficients on the geographical distance variable’ (2000, p. 17). However, such analysis does not specify whether there are sectoral or spatial structure differences. There can be a significant distinction between a general confirmation of a hypothesis that knowledge is localized and a finding that there is a specific structure to the linkages or that the importance of inter-regional links are differentiated across sectors. Thus, more research on knowledge linking patterns

11 Saxenian identifies the wagon wheel bar in California as a key location for the flow of informal information in the early semiconductor industry (1994: 32).

52

3

Innovative Regions, Clusters and Milieux

together with more precision in the language used to describe extra-regional linkages would be a welcome development. This is particularly necessary, as the latest clusters research reveals that firms in clusters do not necessarily preference local knowledge over national or international knowledge (see for example Simmie 2004, and Wolfe and Gertler 2004).

3.5

Cluster Boundaries and Extra-territorial Linkages

When the prevailing paradigm for geography-based analysis of innovation systems is itself questioned,12 it rapidly becomes apparent that the perspective on such systems is that they are ‘atolls’ of activity. Many of the studies are designed to show why proximity matters. The methodologies and the generality of the conclusions focus attention on the location of agglomerations and the degree to which knowledge spills over borders. This intent for research specifically excludes the consideration of multi-spatial structures. This presentation of ‘islands’ (Bunnell and Coe 2001) of production and innovation sets them adrift from a more transparent presentation of the role of location. Maskell and Lorenzen observe that cluster studies are about co-location (note that there is no mention of how such clusters are linked to elsewhere): ‘There are many competing schools of thought concerned with industrial clusters but they all agree that this real life phenomenon has to do with the co-localisation of separate economic entities, which are in some sense related but not joined together by any common ownership or management. In spite of this basic accord, no general understanding has yet emerged regarding the paramount reason why the separate entities became co-localized in the first place, what has made them stick together, what the effects maybe, and – at an even more basic level – why this matters at all’ (emphasis added 2004, p. 991).

This attention to the characteristics of places corresponds with the position taken in Chap. 2. The overwhelming focus is on endogenous dynamics for growth and development which – defines the study of systems of innovation. Learning, knowledge, research and innovation are all argued to have built in incentives for accumulation trajectories and limited spatial knowledge spillovers which preferences some places over others. Therefore, clusters with ill-defined industrial and geographic boundaries and regions (often with politico-administrative borders) are both typically perceived as located within national borders. Only the analysis of Saxenian and Hsu (2001) could be found that identifies cluster analysis as overly focusing on isolated locations and not the linkages between places. The last analytical section of this chapter examines studies that, in one way or another, can be shown to be related to borders. The first class of research retains the

12

A recent review (Hofe and Chen 2006) of the various literatures with an analytical focus on defining relations within clusters does not even raise the issue of cluster boundary definitions. Seemingly, while the methodologies and approaches for defining the existence and nature of a cluster can be critiqued there are ‘taken for granted’ assumptions about jurisdictional boundaries.

3.5

Cluster Boundaries and Extra-territorial Linkages

53

geographic dimension (proximity) of clustering while loosening another dimension (national borders). The studies of contiguous regions (regions that reside on both sides of a border) reveal that borders are not unimportant for the structuring of economic activity. The second class of studies considered here are those that loosen both cluster dimensions of proximity and border boundaries by investigating the conduct of multinational enterprises.

3.5.1

Contiguous Regions

It is argued throughout this book that neo-Schumpeterian researchers have placed undue emphasis on the nation state as the overriding spatial dimensions of innovation, without considering enough of the external influences and interdependencies. It is not the argument here that either nations or their borders do not matter. As Anderson and Dowd suggest, borders are a critical element of social development: ‘Borders, states and societies are mutually formative – borders shape what they contain and are shaped by them – but border research undermines lazy assumptions that ‘state’ and ‘society’, ‘state’ and ‘nation’ or ‘state’ and ‘governance are co-terminus. Instead of becoming redundant in a borderless world, increased differentiation, complexity and contradictions of political borders make border research more important and more revealing of wider social change’ (1999, pp. 602–603).

The growing interest in the role of borders in shaping development is seen through, for example, socio-political implications (Paasi 1999), Scott on crossborder institutions (1999) and Kraetke (1999) on the influence of borders on regional economic development for firms on both sides of a border. Although crossborder activities could be, conceptually, a label for any range of distances beyond a particular border, it is a term that typically refers to contiguous physical spaces that are divided by political boundaries (Silvers 2000). More rarely, the term has been used when borders occur in combination with physical environment discontinuities such as the English Channel (Church and Reid 1999). The economic geography of neighbouring places13 does seem particularly interesting. The ‘innovative activity in a local system is positively influenced by the level of innovativeness of contiguous systems’ according to the study of Paci and Usai (1999, p. 18) which is based on patent applications to the European Patent Office in 85 branches of industry for 784 municipal regions in Italy. Paci and Usai also note that ‘technological spillovers are not spatially unbounded – since they actually die out with increasing distances from the area considered’ (1999, p. 18). This conclusion is not dissimilar to the findings on the spatial structure of international trade (discussed in greater length in Chap. 4), which indicates that trade is often geographically biased towards closer larger markets.

13

Interestingly, the European Commission would appear to want to encourage cross-border clusters (2002: 32) but the supporting analysis of the phenomenon remains slender.

54

3

Innovative Regions, Clusters and Milieux

The evidence on the geographic dispersal of technology spillovers is also helpful for understanding the operations of large cross-border regional clusters such as the Detroit-Windsor auto corridor. Canada’s auto cluster is just across the river from one of the USA’s prime locations for car manufacturing (see Fitzgibbons et al. 2003). Such an example of a cross border industrial agglomeration is not totally unique but not all are as successful. Hassink et al. (1994) found that technological cooperation across the nation state borders of The Netherlands, Belgium and Germany, in the so-called Maas-Rhine region, was lower than expected. They attributed this to nature of the institutional arrangements which they argued played an important part in encouraging the development of contiguous clusters. Elsewhere, the Øresund Region which covers the territory from Copenhagen (Denmark) to Malmo (Sweden), is being promoted as a possible emerging crossborder region, largely because of the construction of a very large transport system (bridge and tunnel) between the two cities. Bucken-Knapp thinks that in the medium term future, rather than the development of a unified region, it is more likely that initially Copenhagen based workers might relocate to the cheaper Malmo area. Inturn and in time, this might then entice businesses to set up in Malmo. Whether the evolution suggested by Bucken-Knapp would lead to one Øresund region or two linked regions is not addressed and is a question the scale at which one might want to undertake analysis.

3.5.2

Clusters Beyond Proximity

The next step beyond contiguous regions is conceiving of space as a series of places which are linked by differential ties, but this is rare within the literature on clustering. The literature on economic geography and clusters, in particular has only moved in small steps towards a framework that can integrate an acceptance of both the importance of spatial innovation systems (see Martin and Sunley 2003) and the importance of distance centres of production and knowledge creation. It is clear, however, that this is a direction which is beginning to emerge in the literature (see in particular Wolfe and Gertler 2004). A few different perspectives on dispersed production systems exist. The leading edge of research suggests that external linkages are important (Wolfe and Gertler 2004). Some of this research highlights the significance of external relationships for knowledge sharing and differentiates between local buzz (tacit knowledge) and global pipelines (deliberately developed relationships for exchanging tacit knowledge), described by Bathelt et al. (2004), amongst others. Although I would disagree with Faulconbridge’s view that the concept of local places as ‘knowledge “nodes in global networks” (Amin and Thrift, 1992) is now somewhat banal’ (2007, p. 1635) I am sympathetic to the finding that: ‘important similarities are revealed in the way ‘buzz’ is produced in long and short, local and global learning networks. Indeed, this in-depth examination of the practices of learning,

3.5

Cluster Boundaries and Extra-territorial Linkages

55

something often missing in existing studies of clusters and translocal learning networks (Bathelt et al., 2004), reveals that similar architectures of learning exist between individuals in close and less close physical proximity with relational proximity being the defining factor in the success of learning’ (2007, p. 1652).

Alternatively, Markusen (1996) describes three types of clusters with external relations: ‘Marshallian’, ‘hub and spoke’ and ‘satellite districts’. Marshallian industrial districts have a strong set of internal capabilities with a mix of different sized businesses, whilst ‘hub and spoke’ regions are dominated by a few large organisations that are supplied by many small businesses that are often but not necessarily located in the region of interest. In contrast, ‘satellite districts’ are dominated by large businesses that are focussed on supplying business units that are external to the region, which results (in Markusen’s view), in the lowest range of regional capabilities. These typologies, although based on empirical analysis, are only identified with specific locations in the most generic terms. This is unlike Guerrieri and Pietrobelli (2000) who applied Markusen’s Marshallian districts typology to their work on regions in Italy. Their focus on the ability of systems and actors to evolve and technologically upgrade led them to conclude that firms working solely within a cluster may not be able to access relevant technological knowledge. They argue that businesses not within global production networks (discussed in Chap. 4) will not have access to important commercial and technological knowledge. Other researchers also question whether clusters are based in proximity economics at all. ‘Innovation relationships’ in the German mechanical engineering industry ‘are far less spatially restricted than generally assumed’. In many cases, access to interregional contact networks is much more important according to Grotz and Braun (1997, p. 555) – especially for technology-intensive firms. The policy implications they draw is that cluster or regional policies that result in tight integration is not necessarily always the best result. Buyers and sellers are also often located distant from core centres of attention. Suppliers to innovative firms in Amsterdam, Milan and Stuttgart were more local than extraterritorial yet suppliers to innovative firms in the two global cities of London and Paris, were more internationalised. In all five cities, customer relations were more internationalised than, supplier relationships, according to Simmie et al. (2002) and Simmie (2002). These results are not generally surprising given the existing research on world cities and the business service firms that are located within them (see Chap. 4). The literature on cities reports that the top tier are more linked to cities in other countries than their own, while other tiers of cities are more focussed on intra-national markets. Later work by Simmie (2004, p. 1111) on firms in the UK led him to argue the following: ‘contrary to the local clustering hypotheses, market-leading innovative firms seem to be more a part of an internationally distributed system of innovation. Their clients and customers are located around the advanced economies particularly in Europe and the US’.

The flip side of research, that identifies the external relations of firms based in particular places, is the analysis of the rationale behind multi-national enterprises establishing themselves in specific locations. Approximately half of all majority

56

3

Innovative Regions, Clusters and Milieux

owned foreign affiliates (MOFAs) establishing in Sweden located their operations in just three competitive host clusters (Ivarsson 1999). Once embedded in these clusters, the foreign affiliates oriented their production towards export markets. The other half of the newly established MOFAs did not base themselves in clusters and were primarily selling to Sweden’s domestic market. Both types of MOFAs relied on local material inputs to about the same extent and both sets of MOFAs cooperated (to a small degree) on technology development predominantly with sister firms in their own business group. The results of Ivarsson would indicate that the export-oriented firms were targeting specific advantages that could be gained from accessing resources (human capital, etc.) embedded within Swedish clusters. Therefore, although, formally tapping into technology in local clusters may be difficult, a conclusion of other research (see for example Patel 1997), there are economic benefits of being inside clusters. This is an important conclusion. In contrast to the forgoing research, it should be noted here that there are prominent cases where omitting external linkages from consideration has resulted in clearly deficient analysis. The emergence of Ireland as a high-income country via a sustained period of high growth is one of the most important changes in a country’s industrial trajectory in the last quarter of a century. There are various analyses of the success of Ireland but there is an obviously inadequate consideration of international links. The research of Green et al. (2001) includes a discussion of the relative scale of MNE investment in the Irish ICT sector with the claim that this activity has become ‘embedded’ in the national structures through the rise of local firms – and thus the formation of local clusters. However, there is only a passing mention to the scale of imports required by the Irish cluster. In Chap. 9, it is revealed that more than a third of output in the electronics national-macro cluster was contributed by component imports. In a similar article on the role of investment and R&D, O’ Sullivan attributes Ireland’s success in attracting foreign investment to the profitability of plants based in Ireland which in turn are founded on wage rate advantages and high productivity. The extent to which this “production base’ for Europe (2000, p. 278) is in fact supported by significant volumes of imports is again not discussed. The emphasis on local factors seen in the research of Green et al. (2001) and O’Sullivan is relatively commonplace in the literature on innovation. In the article by Hicks et al. (2001) on the citations of local academics within computer patents in the USA, identified earlier, has an important twist. The article emphasises the relative importance of the local connection between inventors and local research. However, it notes only in passing that in absolute terms the corporate patents overwhelmingly cited14 academic research conducted in just two US States (California and Massachusetts). Thus, the cluster linkages to the major American centres were downplayed in favour of emphasising a relative local preference.

14

31 per cent of all references.

3.6

Multi-spatial Production and Innovation Spaces

3.6

57

Multi-spatial Production and Innovation Spaces

It is clear from the evidence presented in this chapter that the phenomenon of industrial clustering is significant, even if the questions of appropriate spatial scales (Malmberg and Maskell 2001) or which industries (Steinle and Schiele 2002 or Audretsch and Feldman 1996) remain contested. From all the evidence compiled for the current chapter it is argued that a multi-spatial perspective on innovation systems is largely missing from the sub-national systems of innovation analyses, just as it was when the focus was explicitly on national systems. Increasingly, however, there is emerging fragmentary research that reveals that clusters are linked to national and international sources of technology. Wolfe and Gertler comment: ‘The picture already emerging from our study departs substantially from the received wisdom – most notably concerning the alleged importance of a strong local customer base and strong local competition in spurring the emergence and evolution of dynamic, knowledgebased clusters. … In particular, it appears that a large component of the knowledge inputs to local production – at least in certain sectors – is drawn from well outside the region’ (2004, p. 1090).

This acknowledgment is welcome but there is still very little information on the scale, spatial structure of such external connections or their characteristics. Despite this, the literature on territorial (urban, regional or clustering) innovation systems is more productive in examining the role, scale and spatial configuration of interdependencies than the national level studies. The available research literature tends to suggest the following conclusions: • Productive systems tend to agglomerate in particular locations; • There are many names for these agglomerations, but ‘clusters’ has become the most frequently used term to designate a combination of both geography and sectoral emphasis; • Clusters are seen as having both untraded (knowledge) and traded dimensions (industrial supplies); • Untraded interdependencies are seen as the diffusion of predominantly tacit knowledge within a limited geographic area (so proximity is important for knowledge flows); • Long-lived user-producer relations tend to encourage innovation and are a prerequisite for innovation; • Such user-producer relations are thought to be stronger is particular localised spatial identities (regions or nations); • Clusters have been identified using input-output analysis to reveal highly linked industry groupings; • Cluster studies tend to focus on the internal dynamics of particular geographic entities; and • Crucially, for the argument here, those studies that have specifically focussed on the question of the external connectedness of regions have mostly suggested that the linkages were stronger than a proximity-based hypothesis would assume.

58

3

Innovative Regions, Clusters and Milieux

On the basis of the research reported here, it can be argued that the cluster literature provides enough evidence to support an argument that proximity does in some way influence outcomes and that firms do tend to co-locate whether competing, supplying or consuming. However, there is no theoretical framework for considering clusters within a multi-spatial agenda. Research that has highlighted the importance of imported knowledge or supplies do not go the next step and investigate the spatial structure of those extra-regional linkages. The following chapter examines the linked component of the multi-spatial framework. The chapter reports on current research findings and approaches to trade, both intra-nationally and especially internationally. It establishes the usefulness of current methodologies for trade analysis and research findings on the spatial structure of global trade. To do this, the chapter examines the literature on bilateral trade relationships (specialisation, development, duration, and strength), the impact of borders on trade, global production networks, world city trading networks and multi-regional input-output research. The chapter also brings together the evidence from chapters two, three and four to present an overall summary of the arguments for a linked clusters research agenda that concentrates attention on the role, scale and spatial structure of linkages in innovation systems beyond national borders or regional boundaries.

Chapter 4

Beyond Borders: Trade and Networks

‘We need to better understand the various roles that local agglomerations play within spatially extensive value chains and begin to map the activities that tend to concentrate in particular places even as the geographic ‘footprint’ of linked economic activity expands. It is the linkages mechanisms, between firms and between places that especially deserve more of our research attention’ (Sturgeon 2003, p. 200).

4.1 Traded Interdependencies Beyond Borders In Chaps. 2 and 3 it was shown why neo-Schumpeterian researchers focusing on the operations of systems of innovation have emphasised both the nationally conditioned factors fostering innovation and those elements of the processes of innovation that drive proximity dynamics. It was also shown however, that the innovation literature has been overwhelmingly captured by a paradigm of endogenous capability. Analysis of the literature reveals that it undervalues the significance of the external environment by largely ignoring the scale of extra-cluster linkages and, in particular, being disinterested in the spatial structure of interdependencies, an argument also made by Bunnell and Coe (2001) although they go little further. This chapter explores how a number of different research traditions have investigated both trans-border activities and multi-spatial systems, whether they are connected across intra-national regions or countries. This analysis facilitates a presentation of the linkage aspects of the linked clustering framework suggested here. In doing so, this chapter traverses some diverse academic traditions. Bilateral trade analysis (both neoclassical and neo-Schumpeterian) is considered, especially noting the empirical evidence on the role of international borders and regional boundaries in strongly influencing the strength of trade. The evidence for international inter-connectedness is considered from a broad spectrum of analytical perspectives, including trade theory, global production networks, global commodity chains, global and world cities, and production fragmentation. The emphasis in this chapter is to scan the literature for empirical clues on the scale and spatial structure of trans-boundary interdependencies. The breadth of literature covered is designed to ensure as far as possible that the suggested B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_4, © Springer-Verlag Berlin Heidelberg 2009

59

60

4

Beyond Borders: Trade and Networks

Fig. 4.1 Innovation, trade and international networks literatures

multi-spatial innovation systems framework is not simply a re-statement of research already presented elsewhere just wrapped in a new guise. Figure 4.1 translates these foci into the format of the literature diagram. The important feature of data on industrial production over the last 20 years of the twentieth Century is that while advanced economies have maintained their specialisations, the production of technologically complex products has emerged in newly industrialising countries. No longer is it just production of low cost per unit products such as textiles, clothing and footwear that is conducted in developing countries. Wixted (2005) presents a series of bilateral trade charts that ‘map’ the evolution of industry exports and destination markets for East Asian economies, revealing the change from ‘low technology’ industries to a concentrated specialisation in electronics (primarily focused on the US market). Various research traditions have met (or not) the challenge of understanding this change. Each have there own emphasis, but rarely do they combine trade with spatial and innovation elements.

4.2 Trade, Borders and the Sourcing of Products There are two dimensions to trade analysis, regardless of whether the goods and services are crossing international borders or simply politico-administrative boundaries within a nation state (e.g. States and Provinces). These are the ‘what is traded’, in relation to other products (specialisation) and the ‘with whom it is traded’ (bilateral relations) questions. Both topics are included here to present an understanding of the evolutionary patterns of trade and how those patterns are related to patterns of technological innovation.

4.2 Trade, Borders and the Sourcing of Products

4.2.1

61

Trade Specialisation Patterns

A key component of the development of a multi-spatial innovation systems perspective is the need for more analysis of the significance of external interdependencies for regions and countries. To begin, it is necessary to start with, the neo-classical trade theory of comparative advantage, which is still the most widely accepted paradigm for understanding international economic relations. Comparative advantage is not just the dominant perspective on why countries trade what they do, but is the paradigm against which the neo-Schumpeterians have developed their research agenda. Comparative advantage exists where countries trade together in products where they have a relative advantage. As Leamer and Levinsohn explain, ‘international economics is concerned with the trade of a country vis-à-vis the world and has little to say about trading partners’ (1995, p. 1385). The current incarnation of the theory is the result of various changes as noted by Krugman and Obstfeld: ‘The Ricardian model. Production possibilities are determined by the allocation of a single resource, labor, between sectors. The model conveys the essential idea of comparative advantage but does not allow us to talk about the distribution of income. The specific factors model. While labor can move freely between sectors, there are other factors specific to particular industries. This model is ideal for understanding income distribution but awkward for discussing the pattern of trade. The Heckscher–Ohlin model. Multiple factors of production can move between sectors. This is a harder model to work with than the first two but conveys a deeper understanding of how resources may drive trade patterns’ (1997, p. 93).

Supporters of comparative advantage theory still accept its applicability (see Leamer and Levinsohn 1995 or Davis and Weinstein 1996) and strongly advocate its continued accuracy. However, during the last 20 years or so the rapid global growth of intra-industry trade between many pairs of countries has posed an increasing challenge to the theory. Intra-industry trade (IIT) is the export and import of products from the same industry (defined by statistical codes) between two countries. It is probably the most cited criticism of neo-classical trade theory. The argument is that pairs of economies trading in their respective industries of comparative advantage should be specialising in different industries, not the same ones. This problem is, however, according to Fontagne and Freudenberg only a statistical artefact. They have shown that at a sufficiently disaggregated level intraindustry trade ceases to exist and is replaced by highly differentiated trade. This is more in accordance with traditional economic theory. ‘Using a dataset embodying data flows of 11 European countries facing 10 partners for around 10,000 products, the methodology emphasises that the recent increase in IIT in Europe is entirely due to a trade in vertically differentiated products’ (1997, p. 8).

Thus, countries specialise in different products, either through product variation or through product quality. These results go some way towards reconciling comparative advantage theory with the empirical evidence, but of more concern to researchers of the economics of innovation is the comparative advantage model of the division of labour. The primary

62

4

Beyond Borders: Trade and Networks

target of these researchers is the unrealistic assumptions of knowledge generation and diffusion. The EITC field is keen to develop a framework that replaces comparative with technological advantage. It might be possible to alter neoclassical trade theory’s foundations of common technology and the relative prices of the factors of production (land, labour and capital) accounting for the differences in trade or extending these comparative advantage models to include the factor endowment of ‘knowledge’,1 but this problematic. Instead, EITC research advances an endogenous view of differentiated technological capability. One of these key endogenous variables is that there is a link between areas of knowledge creation, trade specialisation and economic growth. In the first instance, the evidence for trade specialisation is similar to the findings on industry specialisation reported in Chap. 2. There is a high degree of stability2 in trade patterns, although technological change can disrupt these configurations, as Amendola et al. (1998, p. 152) note, ‘there are sectors in which technological and trade advantages (disadvantages) have remained stable and sectors in which comparative advantage (disadvantages) have changed substantially during the longer time span considered (1967–87 for patenting and 1970–87 for trade)’. In this analysis of OECD countries, three groups emerged: The first group includes countries that were technological leaders and which had developed strong links between technological and trade specialisation. The second group were countries with weaker connections and thirdly, small countries that had a strong association between trade and technological specialisations. For many high-income countries, therefore, investment in technology co-evolves with trade development. Although countries are not typically de-specialising (Amendola et al. 1998, p. 147), country competitiveness as analysed by Guerrieri and Milana (1998) has changed. Between 1970 and 1992, Japan improved its position, while European countries suffered losses in their competitiveness. Fagerberg (1996) attributes long run competitiveness to the level of R&D and innovation, the size of the domestic market for very high technology industries (e.g. aerospace) and the benefits that accrue from overall national R&D expenditure to a wide number of industries through indirect knowledge spillovers. What is less clear is how trade patterns evolve within a neoSchumpeterian framework. Amendola et al. (1998) argue there is more interest in the variables associated with trade patterns than the trade patterns themselves, a position that is still persuasive. It is, therefore, worth recognising that both factor endowments and endogenous innovation are important for the continuance of trade specialisations, although this is more than would be recognised in the neo-classical literature.3 1 See Verspagen and Wakelin (1997). Of course, the entire premise of knowledge endowment is disputed by neo-Schumpeterian authors that argue that; human capital, R&D and knowledge all arise from systematic investment. See Davis and Reeve 1997 for an endogenous model of human capital. 2 Observed by both Guerrieri 1999 and Amendola et al. 1998. 3 See for example the references in Davis’ paper ‘Understanding International Trade Patterns: Advances of the 1990s (2000).

4.2 Trade, Borders and the Sourcing of Products

4.2.2

63

Bilateral Trade Relations

If the first dimension of trade analysis is the ‘what of trade’ (specialisation patterns), the second dimension is the spatial structure of linkages. Bilateral trade research most directly addresses the issue of the destination of trade and the reasons behind the choice of trade partner. This field, like that of trade specialisation, is also divided between neo-classical and neo-Schumpeterian perspectives. Of particular concern here is discovering whether the linkages across geographies exhibit a nodal structure in an extended version of the characteristics of social relations and knowledge diffusion apparent in the networked nature of geographically concentrated clusters.

4.2.2.1 The Contours of Bilateral Trade: An Overview of Trade Structures It is worthwhile starting the discussions of bilateral trade patterns by presenting some baseline on manufacturing trade between OECD countries. Figure 4.2 reveals the specificity of trade relationship composition (at a highly aggregated level both sectorally and in trade volume). Using the OECD Bilateral Trade Database (1997), a matrix of manufacturing trade was constructed for 1994 with a threshold for the exporting country in Fig. 4.2 of 2% of GDP, and in Fig. 4.3, a threshold of 1% of GDP was adopted. The choice of either importer or exporter is critical for the overall visual outcome of the charts. Exports are a standard economic interest, because it is seen as a measure of the competitiveness of economies and the results for smaller countries are more easily interpreted. If imports were the basis of analysis, imports by larger economies from smaller ones would be harder to detect against the volume of other economies.

Fig. 4.2 Manufacturing export linkages greater than 2% GDP (1994) Source: Data from OECD (1997) and GDP data from IMF (2000)

4

Fig. 4.3 Manufacturing export linkages greater than 1% GDP (1994). Source: Data from OECD (1997) and GDP data from IMF (2000)

64 Beyond Borders: Trade and Networks

4.2 Trade, Borders and the Sourcing of Products

65

The high statistical threshold for trade linkages was adopted because at very low levels of the value of trade, especially for the entire manufacturing sector, most OECD countries would be seen to be trading with most other OECD countries. Figure 4.2 reveals that at these high levels of trade concentration it is the smaller economies with very high levels of exports (as a percentage of the domestic economy) to the larger economies of the USA, UK and, in particular, Germany which stands out. Trade source countries tend to have one or two significant destination countries, a few have more. Some of the significant links are: • • • • • • • •

Austria exporting to Germany; Canada exporting to the USA; Denmark exporting to Germany; Iceland exporting to Japan and the Netherlands; Ireland exporting to France, Germany and the UK; The Netherlands exporting to Belgium–Luxembourg, France, Germany and the UK; New Zealand exporting to Australia and the UK; and Switzerland exporting to Germany.

Australia is an interesting case because although it is a mid sized economy it does not export manufactured products to any country at or beyond the 2% of GDP threshold. Naturally, as the share of GDP is decreased more links are revealed. For 1% of GDP the destination countries of Germany, the UK, the USA and France are again the primary international markets. Some smaller countries such as Finland, Ireland and New Zealand have a significant number of links whilst the mid-sized economies of Canada and Australia have only one trade link (USA and Japan respectively). Thus, trade patterns tend to be highly concentrated. At greater levels of aggregation, most economies are focussed on very few destination markets. As the results presented in Chaps. 6 through 9 mainly focus on imports these results on export patterns help us to understand the overall patterns of global interactions.

4.2.2.2

Long Lived Relationships: Exports and Imports

Not only is the web of relationships for particular sets of partners relatively concentrated at greater levels of trade volume, the structure of relationships for OECD (and for many) countries only changes very slowly. From a macro statistical perspective, Drysdale and Garnaut (1994, p. 29) identify imperial trading blocs as an initial determinant of trading partner preference. As these blocs disintegrated, there was a very gradual movement from these linkages towards ‘relative distance’ (larger closer markets) as an explanatory variable of trade partner choice. Bilateral relationships do change and those revealed in Fig. 4.2 are gradually de-specialising – each country trading with a greater number of partner countries. Laursen (1998a, b) suggests that this ‘country-wise’ specialisation trend is generally holding but the sectoral trade specialisation varies between various machinery

66

4

Beyond Borders: Trade and Networks

industries (transport, and energy, etc.) which are not greatly internationally mobile and electronics based industries that are more mobile. Analysis by Laursen vis-à-vis technological specialisation trajectories reveals a more country dependent arrangement. Combining technology and bilateral relationship data reveals that for many sectors, technology gaps partially explain trade partner choice (see for example Wakelin 1998). These lessons from history need to be tempered with the knowledge of the rapid rise of China as a source of production. In contrast to previous patterns, the rise of China has been shifting production around relatively quickly. Meng et al. (2006) reveal that Asian production is developing a tri-polar system based on the USA, Japan and increasingly China. While the pattern of exports is important, as the latter chapters in the book attempt to show, it also matters a great deal, perhaps more so, what countries need to make those products. However, there has been less consideration of the role of imports compared to exports. Export specialisation has been used to test a wide variety of hypotheses ranging from the significance of investment in technology, the degree of path dependency and cumulative causation in trade development, and the role of national innovation systems. Imports across nation state borders are seen as a negative, however (see Archibugi and Iammarino 1999). Although inter-regional trade is rarely considered – traded interdependencies within a region are used as an indicator of cluster strength, business networking, and corporate technological specialisation. Only a few studies of imports and technologies could be identified for the current study. Notably Laursen and Meliciani (2002) conducted an extensive econometric analysis of the value of imports for international competitiveness. They found that, in general, international technological spillovers were small. Their methodology attempted to account for the individual country sources of technology by incorporating a calculation of the R&D intensities of trade partners, but the model developed needed to rely upon the sum of imports. In this way, it is not possible to use their results to explain whether imports from particular countries were important. In general, the evidence from the modelling conducted for the present book would support a general conclusion that imports still play a small part in competitiveness. However, on the basis of intermediate goods transactions alone (presented in later chapters), without any calculation of embodied R&D, there would appear to be cases for specific combinations of bilateral trade relations for specific industries where imports are important. In a similar vein, a recent study by Srholec (2007) directly questions the degree to which export statistics reflect indigenous capability. He notes that although R&D data is available for many advanced economies, it is not yet available for many rapidly developing economies, particularly in East Asia. By using an analysis combining exports and imports he concludes that for many East Asian economies, their high technology profile is somewhat dependent on imports suggesting that their endogenous capability is less than their export indicators alone would suggest. He notes that Korea, Taiwan and Singapore are exceptions. The analysis presented in latter chapters of this book adopts the methodology of integrating the data on imports into the value of production.

4.2 Trade, Borders and the Sourcing of Products

4.2.2.3

67

Social and Economic Factors in Business-to-Business Networks

These statistical analyses of trade direction, as valuable as they are, don’t assist with developing an understanding of what is underlying the creation of trade relationships – one firm at a time. One of the important ingredients in cluster development (as shown in Chap. 3 above) is the social relations that exist in particular localities and assist with the flow of knowledge. Relationships between producers and users can also develop to the point where they promote the introduction of innovations. The argument in the neo-Schumpeterian literature often implicitly and, on occasions, explicitly assumes that a dense network of social relations are based in regional proximity and do not extend across national borders. Inter-regional or international trade under this theory only builds traded interdependencies (with embodied technology) and not tacit knowledge flows or user-producer relations. To question therefore, as the current work does (in part), that local interdependencies need not be fundamentally different to distant interdependencies, the social embeddedness of trade needs to be established. One of the clearest pieces of evidence on the knowledge density of international trade, the value of business networks and the role of person-to-person contacts, is provided by Drysdale and Garnaut, who describe a study of Garnaut’s where he had interviewed company executives), on the development of export markets. ‘The study revealed that the preferences of ultimate users were important in determining bilateral trade patterns in highly differentiated commodities, but not in more homogeneous commodities. For the latter, biases in the trade decision-making processes of companies were of considerable importance. For affiliates of multinational enterprises, which accounted for a substantial part of Australian and Southeast Asian foreign trade, there was a high degree of intra-company trade, so that the location of affiliates was a major determinant of bilateral trade patterns. For independent enterprises, the particular order in which pioneering trading firms searched the international environment for trading opportunities was of considerable importance, especially since many firms relied heavily upon other trading firms in their own country for leads on new markets. Explicit internal constraints on profit maximising behaviour appeared relatively unimportant in the determination of trading patterns’ (1994, p. 29).

Drysdale and Garnaut provide three insights. First, in highly differentiated products users of those products were important for creating and sustaining the relationship. This fits very neatly with the user–producer innovation literature. From this, it is possible to conjecture that in technology based products users know which technologies/products are required to fill their own technology gaps. Second, intra-firm trade is important. The movement of goods and services inside a multi-national’s own existing network could generate a nodal structure (if it accounts for a high percentage of trade) within international trade. Thirdly, the search for business partners is conditioned and structured by the knowledge of where other similar businesses have opened up markets. This would involve the transfer of tacit knowledge at one end of the production chain (sellers) and probably both ends with potential buyers being aware of the reputation and value of products from a particular source location. This dynamic would also influence trade decision making helping to generate a nodal structure. As the cumulative causation of imitating behaviour

68

4

Beyond Borders: Trade and Networks

leads more and more businesses based in one location to seek customers in similar markets, it also builds the structure of relations observed in Fig. 4.2. Interestingly, a research lead from the clusters literature may shed new light on why firms in a particular location may follow other businesses (including potential competitors) into the same markets. Chapter 3 analysed many of the existing threads of research in the clusters literature but it did not highlight one particularly novel approach put forward by Maskell and Lorenzen (2004) on economic agglomerations. It is discussed here because it links well with some literature on international trade networks. Although the profit maximisation function of businesses typically receives the most attention, firms operating within market economies are also attempting to minimise risks. Domestic firms face technological discontinuity, competitor strategies, business cycles and exchange rate fluctuations, while those corporations that export face multiplied risks,4 as there are now two or more markets to be managed with less access to information on potential buyers. The advance of Maskell and Lorenzen is to address how some of these risks can be minimised in a local setting. The local cluster, as these authors describe it, is one type of a number of different kinds of markets. The cluster then, is a ‘market organization that is structured along territorial lines because this enables the building of a set of institutions that are helpful in conducting certain kinds of economic activities’ (2004, p. 1002). Maskell and Lorenzen’s insight is to show that a restricted market in the form of a cluster allows for the development of a selection of producers or users without being forced into an open market situation with the greater risks and less opportunity for technological choice, that situation entails: ‘Industry uncertainty implies that not all industries are characterized by reasonably stable sets of suppliers, customers and products. With high levels of uncertainty it makes little sense for firms to engage in network building with what will soon become yesterday’s partners. Firms finding themselves in such circumstances tend instead to opt for a strategy of being a stakeholder in a cluster. Within a cluster, the structuring of markets usually takes place with the participation of more ‘insider’ firms and on a broader level than if embedded in a business network only. The extended range of ‘insiders’ with their own capabilities and resources allows for experimentation, flexibility and the use of shifting combinations of partners without carrying the full burden of spot-market transaction costs’. (2004, p. 995)

Clustering can thus reduce uncertainty while facilitating experimentation and ipso facto reduces search costs simultaneously. Although Maskell and Lorenzen keep their focus on localised markets, their framework for considering clusters as a market organisation for reducing risks has strong parallels in the work of Rauch on international trade networks, although the similarity may not be immediately obvious. Rauch notes that trade occurs in networks not markets. ‘It is well known that very few manufactured (as opposed to primary) commodities are traded on organized exchanges. Instead, connections between sellers and buyers are made through a search process that because of its costliness does not proceed until the best match

4

See for example the UK government related website http://www.sitpro.org.uk/trade/managingrisk.html (accessed 15 Jan 2008)

4.2 Trade, Borders and the Sourcing of Products

69

is achieved. This search is strongly conditioned by proximity and pre-existing ‘ties’ and results in trading networks rather than “markets”’ (2001, pp. 7–8).

Maskell and Lorenzen understand the networks in clusters as a form of ‘market’ but Rauch argues that the lack of ‘spot-markets’ and the creation of ‘networks’ should be interpreted as the existence of non-markets. In fact, in Maskell and Lorenzen’s terminology, markets exist when ‘some sets of characteristics are so common that they represent a specific market organization or market form’ (2004, pp. 1001–1002), whereas to Rauch, a market is an openly competitive system of contract purchasing based on product qualities, price and other attributes. Thus, rather than simply limiting the definition of markets to be perfectly competitive markets, the arguments used by both are basically in alignment. Businesses need to reduce the costs of open competition markets and create a new economic space that is structured to trade off flexibility and lowest costs against less uncertainty to form what could be described as restricted markets (clusters or trade networks). Rauch notes that: ‘numerous statistical and case studies provide evidence that trans-national business and social networks promote international trade by alleviating problems of contract enforcement and providing information about trading opportunities (2001, p. 1200).

Put succinctly, Rauch comments ‘[social] transnational networks can help to overcome informal barriers to international trade’ (2001, p. 1190). These networks in the sense used by Rauch are typically social groups (ethnic or religious groups) or business associates (allied but without formal ownership, e.g. ex-employees of the same company). He limits his analysis (2001) to those networks that were initiated domestically and expanded across borders. Curiously, Wolf (2000, p. 555) provides additional supporting information noting that ‘inward migration appears to be robustly associated with export growth’. Thus, networks whether they be local (clusters) or long chain (international trade) can be understood as a necessary practice to reduce the risks associated with open markets. The operation of business networks is the subject of its own specialised field (industrial marketing)5 that dates back at least 50 years. The core research agenda of industrial marketing has been the creation, nature and continuance of long lasting relationships between industrial suppliers and users of products. Much of the analysis has had a focus on the behavioural elements of network development rather than the purely economic incentives. Cova and Salle (2000) analyse the social dimension of business relationships by examining the rituals of business-tobusiness relationships. Ritter (2000) studied the impact one set of relationships has on the web of network partners developed by a business and Katsikeas and Kaleka (1999) looked at ‘import motivations’. Lastly, although the literature presented in Chaps. 2 and 3 of this book emphasises the necessity of proximity for a link between user – producer relationships to promote innovation and learning, this may not be required. Apparently, buyer–supplier

5 Wilkinson (2001) provides a very readable account of the history of ideas on the development of (industrial) marketing networks and channels creation.

70

4

Beyond Borders: Trade and Networks

networks that extend across international borders can also be channels for technology and knowledge transfer. The industrial marketing literature (see e.g. Thompson et al. 1998) reveals detailed information on the complex nature of buyer–seller communications and the importance of reducing the risks of non-compliance with requirements. Rauch (2001, p. 1197) as well, finds that buyers in developed countries are a major source of information for producers in less developed countries, particularly through instructions and standards. Therefore, trade partner developments are a mixture of social, behavioural and economic factors, which are business practices that are unlikely to change with new communications technologies. In the view of Leamer and Storper ‘the Internet will probably re-enforce the roundaboutness of production and hence the importance of face-to-face contact, though it will also probably make possible greater linkages between different localised clusters at very long distances’ (2001, p. 658). More recently, Storper and Venables (2004) have argued that the changes in the world economy and technological change creates uncertainty and complex co-ordination tasks that require the transmission of a high volume of tacit knowledge, which is most readily done face to face through co-location. The advantages of creating networks of business-to-business relationships, whether such networks are local or global, are that they reduce uncertainty and increase flexibility and learning. The idea that clustering is a means of structuring a market needs to be followed up with more research in particular with reference to the possibility raised here that trade patterns are structured by the same need to reduce risks. Curiously, valuable results on this question are emerging from a growing sub-field within neoclassical economics that is attempting to quantify the economic dimensions of the lines that appear on maps, in other words – borders. The analysis has focused on measuring the effect on the volume of trade of both international and politico-administrative borders. This research has been labelled the ‘border effect puzzle’ (Obstfeld and Rogoff 2000).

4.2.2.4

Borders and Trade: The Border Effect Puzzle

The first major paper on the effect of borders on trade was by McCallum (1995) who examined the effect of the Canada-US border on Canadian inter-provincial trade and provincial trade with 30 American states. Using a basic gravity model of trade, McCallum compared actual trade flows to the modelled trade flows. Gravity models of trade, essentially take account of economic size and distance of markets and are relatively good at accounting for real trade activity. Such models would predict that for a certain transport distance a larger market should receive proportionally more exports. However, McCallum reported that 22 times more trade remained internal to Canada in reality, than was predicted by the gravity trade model. This is a puzzle because within a free trade zone, such as exists in North America, neoclassical trade theory would suggest that the absence of trade barriers should lead to trade flowing freely to new markets. The extent of national product preference revealed in the work of McCallum is difficult to explain. Helliwell (1995) determined that the preference to trade inter-provincially within Canada was

4.2 Trade, Borders and the Sourcing of Products

71

even higher for Quebec than McCallum’s average for Canada as a whole. It appears that later analysis has reduced the estimates for the home bias of Canadian–USA trade to about a ten-fold increase above the gravity model predictions (Wolf 2000). In contrast, Anderson and Wincoop (2001) disagree with the home market hypothesis, suggesting that such large trading biases are likely due to the misspecification of the models. Combes et al. (2003) list the top four reasons for why the literature on international borders maybe wrong. These reasons include; poorly constructed analysis and mis-specified equations and models, formal and informal trade barriers could be impediments and thereby reduce cross border trade, the need for different currencies may be a transaction cost that gets in the way of trade, and it has been suggested that national consumer preference for home products could reduce trade across borders. A few authors have looked at intra-national regions. Wolf’s (2000) analysis of intra-national trade in the USA found a similar level of home state bias for States, a factor of approximately ten over the expected neoclassical outcome. This was similar to the results found for the intra-Canadian trade preference. New research by Combes et al. is also based on intra-national analysis. Covering bilateral trade flows between 94 French regions, for 10 industries and two periods (1978 and 1993) (2003, p. 4) the study focused on the effect of regional jurisdictional borders to remove all four of the ‘likely’ problems from border effect research design. Their finding is that business and social networks affect the degree of cross boundary trade even with an intra-country situation. ‘We have shown that intra-national administrative borders significantly matter in trade patterns inside France with an impact of the same order of magnitude that Wolf (2000) finds for trade inside the United States. … When controlling for both type of networks, a French region is estimated to trade only twice more with itself than with a non-adjacent region of similar size and distance’ (2003, p. 34).

Thus social and business networks, which were determined by Rauch (discussed above) to be an important determinant in the structuring of trade relations and reducing risks can also be shown to account for a significant percentage of the trade patterns that do not fit with gravity flow models – a inter-regional intra-national trade (see Combes et al. 2003). However, it is worth noting in passing that none of the explanations for the border puzzle takes account of knowledge, technology or innovation. This is somewhat disappointing as innovation theory has solid explanations (as has already been discussed) for why proximity matters. Nevertheless, given the importance of business-to-business relations, the following section focuses on value chains.

4.2.3

International Value Chains and Spatial Specialisations

Leamer and Storper (2001) acknowledge that production is already clustered, but note that the global trend is towards increased levels of spatial specialisation and more elaborate divisions of labour. In such an environment, it not surprising that in

72

4

Beyond Borders: Trade and Networks

recent years a number of research fields focussed on the internationalisation of production have emerged. Today, many products are made from multiple components and industrial ingredients that have been transported across the world. The fragmentation of value chains, the analysis of commodity chains and the flow of knowledge and technologies within product networks all reveal something of how supply lines are organised to make a product.

4.2.3.1 The International Fragmentation of Value Chains ‘Value Chain Fragmentation’ analysis is frequently an analysis based in the welfare economic considerations of the winners and losers of moving production offshore. The authors in this area are often interested in the welfare implications of changes in trade patterns (see e.g. Jones and Kierzkowski 2001 and Feenstra 1998), but are also interested in mapping the extent to which change is occurring in intermediate goods production (see Hummels et al. 1998 or Ando 2006). The approach relies heavily on the neo-classical trade theory of specialisation (Arndt 1998) rather than on developing an understanding of technological and business environments. Instead, often relying upon modelling, factor endowments and wage competitiveness are considered the main drivers of international competitiveness and economic change. Whilst the empirical analysis of the location of intermediate goods production is useful and has been utilised in Chap. 9 (ICT production), reliance on cost based assumptions of competitiveness while ignoring the consideration of human capital and developing technological capacity in emerging countries impair the analysis. In stark contrast to the traditional neo-classical economic analysis, the ‘Global Commodity Chains’ approach promoted by Gereffi (1998) is more interested in the political economy of relationships. The interest in wages remains, but it also emphases the differential power of purchasing and supplier businesses (supply chain governance). Gereffi comments: ‘What is novel about GCCs is not the spread of economic activity across borders per se, but rather the fact that international production and trade are increasingly organised by industrial and commercial firms involved in strategic decision-making and economic networks at the global level’ (1998, p. 40).

Probably in view of their political importance, Gereffi is especially interested in the GCCs for textiles, clothing and footwear as well as the auto industry. These commodity chains are particularly envisaged in terms of the power relations that evolve when segments of an industrial process are spread between both developing and developed countries. The strength of this analytical tool is its consideration of the strategic positioning of lead players. Its weakness, from the perspective of innovation systems, is that it does not analyse the spatial distribution of technological activities and knowledge.

4.2 Trade, Borders and the Sourcing of Products

4.2.3.2

73

Global Production Networks and Technological Innovation

In contrast to the GCC approach, the ‘Global Production Networks’ (GPN) methodology (see e.g. Zysman et al. 1996, Ernst 2000 and Henderson et al. 2001) does focus on the issues of importance here; production structure, technology development and knowledge transfers. GPN analyses tend to focus on high technology products such as those emerging from the ICT industries or complex product chains such as the auto industry. Ernst (2000) points out that the assembly of an individual computer may involve products from a large number of firms in a number of countries. Another feature of the GPN literature is that it has been focussed on the international business networks operating in Asia, due to the interests of one of the key authors (Ernst). Ernst explicitly adopts a ‘regional division of labour’ perspective (Zysman et al. 1996) acknowledging that it is the combination of geographically centred innovation systems and business networks that are developing and producing the leading edge products. On the basis of his analysis which is mostly at the level of individual corporations, Ernst is in general agreement with the argument that knowledge is geographically immobile, but indicates it is beginning to move more easily, commenting: ‘It is important to emphasize that globalisation should not be reduced to geographic dispersion … [and] does not lead to the wonderland of the ‘borderless world’ where capital, knowledge and resources flow freely around the globe … geographic dispersion has been combined with spatial concentration: much of the recent cross-border extension of manufacturing and services has been concentrated on a handful of specialized local clusters’ (2000, p. 9).

This emphasis on geography, technology and trade is valuable. It would suggest that clusters might indeed be networked through trade into a wider regional production system. Nonetheless, GPN studies focus predominantly on business-to-business network development and the emerging technological capabilities of firms in emerging economies, while the spatial dimension of development is often underplayed. Instead, it is the role of ‘flagship’ corporations,6 which are important developers of new technologies and the builders of global production networks, which attracts attention. Thus, the evidence on the benefits of specific places and the spatial structure of linkages (at the spatial – not corporate level) is more often implied than specifically addressed and is therefore quite fragmentary. Value chain fragmentation, global commodity chains and global production networks research all point to a trend towards increasing levels of production occurring outside advanced economies but none combine the elements of local agglomeration strengths linked via trade to other agglomerations, with an emphasis on the role of knowledge.

6

see Ernst and Kim 2002: 1421 for a diagram.

74

4

4.2.3.3

Beyond Borders: Trade and Networks

Clusters beyond the Nation State, but Still Thinking About Jurisdictions

It has already been commented upon several times in the present work, that the systems of innovation paradigm is framed from the point of view of political entities most frequently – nation states. On the few occasions in which innovation systems are discussed within a multi-national framework, it is not a multi-spatial agenda being proposed. For example, Bergman et al. comment: ‘International trade among cluster members has completely different implications for large vs. small country clusters. A recent study of trade in OECD member countries (Hummels, Rapoport and Yi 1998) shows that vertical trade among international members of a value chain is a much higher proportion of total trade in small vs. large countries. For example, vertical trade is 25% and 42% for Denmark and Netherlands vs. 7% and 14% for the United States and Japan, respectively. The authors consider that these findings reveal a greater likelihood that a cluster’s trading partners are within and therefore responsive to a large home country’s national and regional policies. Paradoxically, however, it also means that supra-national innovation systems (S-NIS) may be essential to sound cluster policies, particularly for small countries. Thus it could well be the case that relevant elements of cluster or innovation policy might logically migrate to the policy frameworks of relevant OECD Member customs unions, such as the EU or NAFTA’ (2001, p. 9) [emphasis added].

There are a number of serious problems with this commentary. First, the conclusion that regions in smaller countries are more dependent on ‘international’ trade than those regions within larger countries is purely an artefact of the definition. By defining trade as international trade7 rather than inter-regional trade, then companies in smaller countries will by definition trade on a more internationalised basis. Secondly, the analyses and the conclusions should be seen as a non-sequitur as they address different issues. The data presented on international trade in the quote reveals nothing of the destination of the trade and therefore no particular reason why there is a need for policy making to shift to the next political level above the nation state (free trade zones and supra-national states such as the EU). The S-NIS perspective has already been analysed in Chap. 2, noting that there has been little analysis of them – even if defined as the EU or NAFTA. In contrast to the argument put forward by Bergman et al. (2001) the break up of value chains would suggest that analysis which follows value chains across borders, yet acknowledging the systemic dimension of technological innovation, is precisely what is required. In some ways, the disjuncture between data and conclusions apparent in this comment by Bergman et al. (2001) highlights the failure of researchers within the neo-Schumpeterian tradition to break free of the constraints of the existing paradigm. In a strikingly similar approach to the proposed S-NIS, Rugman and D’Cruz (1993) criticised Michael Porter’s ‘Competitive Advantage of Nations’ (1990)

7 Trade could be defined as inter-regional, whether or not it crosses an international border. For example do companies in a particular region of Germany or the USA trade with businesses elsewhere in Germany or the USA?

4.2 Trade, Borders and the Sourcing of Products

75

theory of clusters as a framework that is inadequate for small open economies. Because Porter emphasised the importance of domestic demand, Rugman and D’Cruz argue that his analysis understates (for a limited set of economies) the role of leading international supply connections. They develop the idea that as Canada, for example, is strongly linked to the US economy then America represents an extension of the Canadian diamond. This is the basis of their ‘double diamond’ framework. The effect of combining Canada with America is the creation of the North American diamond of business factors. In the Canadian case, the approach seems particularly pertinent but extending this model to countries within the triad8 regions is not necessarily as sensible. There are two criticisms of this double-diamond approach. First, trade is not necessarily triadic in nature. Poon et al. (2000) reveals there are at least four significant trading groups in the world, including the Americas, East Asia, Germany (capturing most of southern and western Europe) and the United Kingdom (which includes some of Africa and Northern Europe). On this analysis, Europe is split between British and German dominated trading systems, and America as a major destination of trade from East Asia is unrepresented. The second reason for being critical of the approach put forward by Rugman and D’Çruz is that even this aggregation of trade zones fails to reveal the complexity of arrangements for any particular industry or country. As an example, New Zealand, a case discussed in passing by Rugman and D’Cruz, but possibly without knowledge of its bilateral trade pattern, has significant exports of manufactures to Australia, Japan, the United Kingdom and the USA. This represents links with members of all three triads, with Australia often not included in any Triad. So what would be the geography of New Zealand’s diamond? It is understandable that policy makers both bureaucratic and political, who are ultimately answerable to a polity, focus on the borders of their responsibility. Researchers, on the other hand should be aware that this perspective potentially misses the opportunity to uncover economic territories that are not necessarily co-determinant with political territories.

4.2.3.4 World Cities: Places, Linkages and Hierarchies Curiously, the fumbling for a description that can adequately handle changes in the world economy within the innovation system literature can be contrasted with the long standing tradition of investigating city systems. Although the research on city development is extensive, there is one aspect of it that is highly relevant to the discussion here. The emergence during the twentieth Century of particular cities as centres for the international agglomeration of business and financial service

8

A broad definition of international regions that include a number of countries: Asia, North America and Europe.

76

4

Beyond Borders: Trade and Networks

activities has given rise to research on the growth dynamics and interdependencies of so called ‘global’ and ‘world’ cities. This literature considers both place and interconnections as integral to the discourse. Beaverstock, for example, has noted the approach of Sassen’s 1991 book. ‘the concept of the global city has emerged because of two inter-related factors: globalization of economic activity, and the organizational structure of the producer service and finance industry itself … the globalization of economic activity, translated as being the shift to services and finance on a global scale, Sassen (1991) believes that these processes have brought “about a renewed importance of major cities as sites for certain types of production, servicing, marketing, and innovation”. In particular, the internationalization of both the producer service sector and financial system has made cities vital centres for the “management and coordination” of economic power in the global economy … the rapid growth, specialization and agglomeration of producer service firms and the organization of the financial industry itself has to some extent been responsible for the formation of global cities. The locational preferences of producer service activities, like, for example, accountancy, advertising and banking, … [means that] producer service firms “obtain agglomeration economies when they locate close to others that are sellers of key inputs or are necessary for joint production of certain service offerings’ (1999, pp. 447–449).

In this view, locations have accumulated advantages and the associated trajectories have often seen places becoming increasingly specialised as the centres of service production. The basis of this hierarchical structuring and spatial specialisation is argued by Sassen (2002) to be facilitated by access to a deep pool of information technology workers and massive broadband capacity. Across time, the accumulation of business service firms and workers can generate economies of scale with particular places becoming increasingly attractive. Given the universal trend amongst highincome economies to move away from manufacturing as a share of domestic income and towards services, particular cities within each economy have become key centres of expertise. Many authors, including Beaverstock et al. (1999), and Smith and Timberlake (2001) have developed ranking methodologies to nominate the networked structure of cities across the globe. At the first tier of cities are places such as New York, London, Tokyo and Paris. Beaverstock et al. (1999) allocated scores to different cities on the basis of the presence of particular global accounting and business service firms. In their tally, Sydney (Australia) ranked alongside San Francisco, Toronto and Zurich, all in their second tier of world cities. Cities can move up or down the ranking and are differentially linked both to their national economies and to the first tier cities. So, ‘Sydney and Toronto have equally gained power in continental sized countries and have taken over functions and market share from what were once the major commercial centres, respectively Melbourne and Montreal’ (Sassen 2002, p. 20). There has been some criticism (e.g. Markusen 1999, p. 875) of the world cities analysis because the concept of the ‘global city’ is indicative of different phenomenon for different authors. It can mean either, a leadership position in terms of international transactions, a concentration of internationally oriented activities or it is a ranking amongst other ‘world cities’ with the implication of some coordinating role of the local financial system. These differences lead to obvious problems in attempts at quantifying the city rankings and any underlying changes. This criticism

4.2 Trade, Borders and the Sourcing of Products

77

does have some validity; Smith and Timberlake (2001), for example, used air travel data as an indicator of inter-connectedness, whilst Beaverstock et al. (1999) used the location of large accounting and consultancy firms to rank the world cities. However, without denying that these concepts remain ill defined,9 the lack of an ideal set of empirical tools does not diminish the observation that different cities are more or less important than others are in either a national or global context. In the analysis of cities, the features of individual places matter, as does the connections between cities. Pred (1977) has made a number of very important conclusions regarding systems of cities. Of note, here is the observation that American cities with dominant positions in the early stages of the industrial revolution retained them until the 1950s and 1960s. Such relative stability profiles have also been a theme of research on innovation systems as noted above. Pred thought that the networks of interdependency between cities were probably the mechanism behind the hierarchical stability commenting that ‘long-term rank stability of large metropolitan complexes … can be most plausibly explained by the tendency of early established major channels of interdependence … to be self-reinforcing’ (1977, pp. 36–37). Such interdependencies were also the probable conduit for the transmission of economic growth between locations. On the basis of Pred, it is possible to construct a line of reasoning on the characteristics of linkages between places. It has been shown earlier that regions and nations have trajectories in economic and technological specialisations which are hard to shift into a new direction. It has also been shown that bilateral relations are stable for long periods. Pred brings these threads together indicating that relationships are a mechanism for positional stability. Interdependencies bind economic islands into economic (or innovation) systems even in a geographically distributed system. The networks of relations establish a set of incentives for a particular trajectory that make it difficult, although not impossible, for any particular location to reposition itself away from an existing trajectory. Therefore, the literature on urban systems contributes many valuable findings. Cities are places of differentiated advantage (clustering), which retain the advantages over a long period of time (specialisation patterns) and exhibit patterns of hierarchical stability (national innovation systems argument). Most importantly, what links these facts together in the urban environment is a system of interdependencies that are both horizontal (the same city tier), and vertical (different tiers) and which lock in the relative hierarchical strengths. The puzzle of location is, however, ‘firms want to locate in an economic centre, but the economic centre is a centre only because many firms are located there. This suggests a process of cumulative causation, in which successive firms entering a location make it more attractive to further firms’ (Venables 1998, p. 4). In contrast, Krugman emphasises that there are economic incentives for centralising as well as movement away from the centre – ‘centrifugal’ processes:

9

A recent article has discussed 100 terms in the urban and regional economic development literature (see Taylor and Lang 2004).

78

4

Beyond Borders: Trade and Networks

‘Many economic activities are markedly concentrated geographically. Yet we do not all live in one big city, nor does the world economy concentrate production of each good in a single location. Obviously there is a tug of war between forces that tend to promote geographic concentration and those that tend to oppose it – between ‘centripetal’ and ‘centrifugal’ forces’ (1998, p. 8).

To some extent, centripetal forces have been addressed in Chaps. 2 and 3 with the analysis of why proximity matters to businesses. The question of centrifugal forces returns us to the issues addressed in this chapter. Comparative advantage, technology gaps, the fragmentation of value chains with ever greater degrees of division of labour specialisation and the establishment of social and business networks together with evidence on the vertical and horizontal spatial structure of linkages between cities all improve our understanding of the movement of production. This is therefore, a useful point to return to the questions, with which the present study began (see Chap. 1) and address the way forward for new research

4.3

Proximity and Multi-Spatial Innovation Systems

This last section of this chapter summarises the evidence on the multi-spatial gap in innovation research, presents a case study of a multi-spatial system and sets the scene for the rest of the book.

4.3.1

Spatial Innovation Systems: NIS and Clusters

In Chap. 1, a number of key goals were set for the analysis of the existing literature on innovation, spatial specialisation and the linkages between places. The first question posed was: 1. Why does the ‘innovation systems’ agenda give primacy to political geography (nation states and regions within nations) over other possible frameworks including economic space, which may lead to more research on extra-territorial links? It indeed does appear from a critical assessment of innovation systems research that the spatial scales of ‘systems’ are limited by politico administrative regional and nation state borders. This seems to be due to: • A predominant emphasis on endogenous technological capability as the key attribute of competitive economies generating wealth; • Technological development processes are often cumulative, as they build on themselves and thus re-enforce particular geographic trajectories; • An overt emphasis on government policy, which is clear in the work of authors such as Lundvall, for example, who notes that ‘instead of looking for clear-cut intellectual origins of the innovation system concept, its main background

4.3

Proximity and Multi-Spatial Innovation Systems

79

should rather be found in the needs of policy makers and students of innovation’ (2002, p. 215); and • There seems to be confusion between the national collection of statistics which makes it simpler to derive national benchmarks (R&D, exports and growth, etc.) and economic structures which may be more regionally based or flow across state/national borders. Despite these limitations, the study of the systemic features of innovation has contributed many valuable insights into the operations of businesses and the dynamics of knowledge within spatial contexts, both national and sub-national. These include the findings that: • Rankings of innovation reveal patterns that are relatively stable across time and so these trajectories have been interpreted as reasons for national innovation system (Patel and Pavitt 1994); • National industrial specialisations are relatively stable across the medium term; • National technological specialisations are relatively stable across time; • National scientific effort is often related to national industrial specialisation patterns; • Technological specialisations are related to trade specialisations for many countries but technological discontinuities can disrupt the pattern; • There has been no extensive movement of the R&D labs of multi-national businesses away from their home country, even if they have shifted their production activities; • Geographic proximity appears to be important both for knowledge transfer and value chain development which contributes to the geographic concentrations of industry [clusters]; • Industrial input-output relationships are relatively stable over the medium term (Verspagen 2002); • National input-output structures have been shown to have some correlation to user–producer relationships [which are critical part of the innovation process] (DeBresson et al. 1998); and • Bilateral export patterns are relatively sticky but European countries patterns are slowly de-specialising (Leamer and Levinsohn 1995 and Laursen 1998a, b). The second goal for the literature analysis was to develop an understanding of the role of interdependencies in economies. The question posed was: 2. What is it about linkages between economic actors (relationships particularly between users and producers) within innovation theory that makes them so important for the development and diffusion of new products and services? The evidence on the role of relationships and interdependencies between economic actors include: • The acquisition of embodied innovation is possible through traded interdependencies; • Business-to-business relationships [user–producer relations] are important for generating innovations;

80

4

Beyond Borders: Trade and Networks

• Generally; local talk and the passing on of tacit problem based knowledge (untraded interdependencies) are seen as the key to regional agglomerations; and • Generally, technology spillovers via national linkages have a higher correlation with export market shares than spillovers through international imports (see Laursen and Meliciani, 2001).

4.3.2

The Multi-Spatial Gap: Cluster Context and Trade Linkages

This evidence, although, revealing that specific locations within the world economy have individual technological strengths that continue across time, also reveals that the innovation systems research agenda has not developed a sophisticated multispatial perspective. It was noted in Chap. 1 that a multi-spatial perspective differs from a benchmarking and comparative approach in so far as it investigates the integration of activities across places and the way performance of one location influences the development of others. The treatment of nations or regions as if they are discrete, implies that that cross border interdependencies are of little significance. This ignores important information on the course of economic history and change at the global scale.10 There can be little dispute that the systems of innovation literature does not seriously address issues of industrial internationalisation11 beyond the level of the firm or attempts to analyse the co-development of clusters. Carlsson notes that ‘Although there is a large literature on the internationalization of economic activity (including R&D) at the corporate level, there are relatively few studies of the degree of internationalization of innovation systems’ (2006, p. 64). Saxenian and Hsu also make a similar contribution when they note: ‘Silicon Valley in California and the Hsinchu–Taipei region of Taiwan are amongst the most frequently cited ‘miracles’ of the information technology era. The dominant accounts of their success treat them in isolation … This paper argues that the dynamism of these regional economies is attributable to their increasing interdependencies’ (Saxenian and Hsu 2001, p. 893 emphasis added).

Two authors that explicitly confront the issue of the interdependency of national innovation systems are Niosi and Bellon (1994) who investigate how national strengths evolve alongside international networks. Interestingly, just as been seen in previous studies, these researchers look to the emergence of supra-national systems

10

Although there have been other periods in history with high levels of international trade (see Williamson 2002), the movement of significant levels of economic activity to new countries is nevertheless an important phemenon (Gereffi and Sturgeon 2004). 11 Similar views have been expressed by Carlsson (2005) and Ernst ‘very little empirical or theoretical research has been done on the way globalisation increases the mobility of innovation across national borders…’ (2000: 1).

4.3

Proximity and Multi-Spatial Innovation Systems

81

rather than the global extension of value chains, noting ‘only one major supra-national system of innovation is presently emerging: the European Economic Community. Canada and the USA show a similar, but less intense, interpenetration of their R&D activities. There seems to be no similar supra-national system emerging that involves Japan’ (1994, p. 195). A different model of analysis, and one which was influential in designing the current study, is Wyckoff’s (1993) work on macro interdependencies between countries and the imported intermediate goods content within industries in OECD countries. The analysis was based on trade and input-output linkages but he was unable to combine the data into a multi-regional input-output model. Wyckoff’s research does not seem to have been followed up with other studies. Wixted has proposed that given the observations in the literature that user–producer relations were often related to innovation and that levels of trade in intermediate goods and services are increasing, it could be possible that international links form ‘supranational clusters’ (2000). Wixted (2005) took this further by arguing from modelling of the industrial interactions of fifteen countries in Europe, and nine OECD economies that some industries clearly exhibited a highly linked structure across national borders. The present work has continued the argument of the earlier two works that the literature on the economics of innovation and technology does not have a highly developed multi-spatial perspective and that such a framework should be defined around the international structure of industrial activity and not political entities. However, it has been consistently argued that the contribution that can be made here is to focus on the role, scale and, with special emphasis, the spatial structure of extra-territorial interdependencies. In particular, the question is whether clusters are nodes within larger production networks or primarily hotspots of economic activity or even relatively (statistically) independent.

4.3.3

Linked Clusters: Specialised Nodes, Globalised Products

As Leamer and Storper (2003) noted, the world economy is evolving towards greater geographic specialisation and greater density of linkages between places. Thus, the third research question raised in Chap. 1 asked: 3. What is known about product and knowledge linkages that extend beyond the borders of particular innovation systems (especially industrial clusters)? While there is a growing literature on international trade which highlights perceived problems with the predominant neoclassical view, it remains entrenched and advocates remain resolute that comparative advantage can explain trade. Leamer and Levinsohn have stated ‘the voluminous and complex literature on the Heckscher– Ohlin models may appear to have left the framework battered and beaten, but nonetheless it remains entirely healthy’ (1995, p. 1375). Importantly, they go one step further, implying that what matters with trade theory are the theoretical and

82

4

Beyond Borders: Trade and Networks

policy consequences rather than empirical observation: ‘gravity models12 are strictly descriptive. They lack a theoretical underpinning so that once the facts are out, it is not clear of what to make of them. In addition they do not link clearly with any issues’ (1995, p. 1387). In contrast, a network based view of trade which has been the analytical emphasis of this chapter, is different and does address the ‘so-what factor’. It appears from the study of the economic effects of borders that trade is more concentrated within a nation than would be compared to what is expected. Social links and business networks have both been shown to be related to this borders puzzle, and are associated with export growth. Company directors (Drysdale and Garnaut 1994) have identified the successful trade patterns of other companies as influencing their own decision-making. Similarly, statistical analysis reveals that social connections between two countries and immigration patterns are mutually associated with trade pattern development (Rauch 2001). This chapter has also raised the possibility that just as local agglomerations of economic activity might be a means for companies to manage market uncertainties (see Maskell and Lorenzen 2004), then businesses might operate within a network of spatially distant cluster nodes to reduce the uncertainties associated with accessing new sources of production and technology. For manufactured products, various frameworks for understanding the operations of multinational enterprises and business networks have been proposed. Global product networks, as one example, are based in specific geographies as well as noting the trade spread internationally. Production and product integration functions can occur in different places and depending upon the technological complexity, it can be a significant number of locations. Technology development functions can be split between network co-ordinators (what Ernst and Kim 2002 might call flagship companies) and local players within the network. Further, there is some evidence, although it is disputed, that MNEs might locate in particular places to access local technological strengths (Ivarsson 1999). Finally, for business and financial services it has been observed that the pattern of city development around the world is structured through spatial strengths and business interdependencies. The role of extra-territorial interdependencies can be summarised as: • • • •

The supply of goods and services (trade based interdependencies); The transfer of information, product standards, etc. (knowledge flows); The transmission of economic growth (on the basis of Pred 1977); and The locking in of hierarchical relativities (suggested by Pred 1977).

12 As noted earlier in this chapter gravity models are generally good at describing trade flows, as typically larger, closer markets are the largest destination for exports – taking note that borders do alter trade patterns.

4.4

Clusters, GPNs or Multi-Spatial Cluster Networks?

83

The scale of extra-territorial linkages (can be summarised as): • Diminish with distance; • Are probably larger than previously indicated in the literature (see recent research by Simmie 2004 and Wolfe and Gertler 2004); • Even knowledge sources for clusters may not necessarily have a particular local bias; and • Based on trade linkages data, borders reduce the level of trade over what would be expected, based on gravity flow models. The spatial structure of extra territorial linkages: • For manufacturing industries, traded interdependencies are often with the nearest largest region; • Bilateral trade patterns are specialised and at high levels of aggregation and are long lasting; • Is influenced by nearby regions, as Beaudry and Breschi (2003) have shown the innovativeness of one region appears to be related to the innovativeness of neighbouring regions; • For manufacturing industries it is unclear whether interdependencies simply extend beyond borders in no particular pattern beyond that which is required for individual businesses or whether they are highly spatially structured so that clusters exist as nodes within linked cluster networks; and • For service industries, the linkages are more likely to be between spatially distant nodes (see for example Gertler and Levitte 2003 and Sassen 2002).

4.4

Clusters, GPNs or Multi-Spatial Cluster Networks?

With so little research conducted to explore the role, scale and spatial structure of extra-cluster linkages, the following case study was developed as a means of teasing out the relevant research issues and to inform the design of the analysis methodology. The case study is of Australia’s emerging technological strength in a specialised sub-set of aerospace components. A number of issues pertinent to the development of this cluster, particularly the role of government policy on foreign investment and purchasing and inter-governmental agreements are not discussed here, although their importance is acknowledged.

4.4.1 Australia’s Aerospace Micro-Cluster As an approach to operationalising the embedded questions in studying the multispatial development of cluster, Table 4.1 was designed. It divides the activities of the cluster (rows) and the geography of the connections (columns).

84

4

Beyond Borders: Trade and Networks

Table 4.1 Linked clustering structure Activities Connections

InterInter-company institutional cluster cluster

National multi-spatial Asia system connections

North America

Europe

R&D Design Component mfg System mfg System integration Services

4.4.1.1 The Products and the Company Hawker De Havilland Australia Pty Ltd. manufactures and sells the upturned wingtips (wingtip fences) as well as other control surfaces necessary for modern commercial jet liners. Recent variations of the Boeing 747 and Airbus aircraft as well as the next generation of passenger airliners require these wingtips for improving aerodynamic performance and fuel efficiency. Fig 4.4 illustrates the wingtip fences.

Fig 4.4 A modern commercial aircraft (747) with wingtip fences. Photograph by Brian Wixted

4.4

Clusters, GPNs or Multi-Spatial Cluster Networks?

85

The aerospace company, which is owned by Boeing, has won a number of contracts, including one to make components for Airbus’s new ‘super jumbo’ A380 jet. ‘Hawker de Havilland will design and manufacture wing tips and vortex control devices for the A380 at its Fishermens Bend plant’ in Melbourne. The new contract will be worth up to AUD$200m (Masanauskas 2003a). In recent years, the company has been able to pick up important contracts with Airbus. ‘Hawker de Havilland already has contracts worth up to $90 million a year with Airbus, including the manufacture of undercarriage doors and floor support structures. The company’s Sydney plant produces wing and landing gear components and will also be involved in the new A380 work’ Masanauskas (2003a).

4.4.1.2 The Global Production Network The wingtips, control surfaces and other aerospace components are sold to both Boeing and Airbus, the two remaining global players in the commercial aircraft industry, even though Boeing owns the company. There is no local market for the company’s advanced composite components so the arguments for local demand being necessary for cluster creation do not hold in this case. However, the current situation does fit within the framework for global production networks. The flagship companies of Airbus and Boeing play the central role in co-ordinating the development and assembly of aircraft. The only major supplier in Australia to these companies is Hawker De Havilland Australia, which against the international corporations building modern commercial airliner components is just a small player. It nevertheless has its own network of specialised suppliers. A GPN framework would also seemingly fit with Hawker de Havilland’s own point of reference. Creedy has reported the managing director as stating that: ‘the company, which employed 1400 people in Sydney and Melbourne, would still position itself as a first-tier supplier for the global aerospace industry and was well placed for the industry’s recovery. It would also continue to play a key role in Boeing’s plan to become a global company’ (2003).

4.4.1.3 A Networked Cluster? However, is this all there is to it – just a case of an international supply chain divided amongst businesses that are linked through ownership or traded interdependencies? The alternative, can this case be interpreted as a multi-spatial system of innovation with specialised nodes within a globalised production system? To argue that Hawker De Havilland Australia is at the core of an Australian node of the global Aerospace industry, it is necessary to be able to show that it is conducting more than just business-to-business sales and that it is more than a GPN. Put differently, if Boeing owns the company, why would Boeing not simply take the knowledge home to the USA? The reason this is not simply a case of a technology-based company investing in research and finding its niche in a world market and nor is it a case of a GPN is because of the involvement of public research and the emerging strength of other companies. The original process technologies for making the composites were

86

4

Beyond Borders: Trade and Networks

developed by the Cooperative Research Centre for Advanced Composite Structures. Australian Cooperative Research Centres (CRCs) are established under a federal Government program which was established in the early 1990s to encourage government laboratories, universities and businesses to co-invest in a single research and development program. Many of the CRCs are incorporated entities with a board and other corporate governance processes and legal rights to the knowledge generated. The contractual arrangements for each CRC and the intellectual property rights of the partners can be very different from one another. This particular CRC13 involves the Commonwealth Scientific and Industrial Research Organisation, researchers (including postgraduate students) from four universities, a range of SMEs, the Federal Government and various collaborators, including Airbus. Hawker de Havilland by being a core member of the CRC, and having contributed both in kind and financial support to the research and development, is able to access the knowledge developed by the CRC. Until the breakthroughs developed within the CRC and now being successfully commercially developed by Hawker de Havilland Australia, the aerospace components (wingtips, etc.) had to be hand made. The expensive process of manually building parts by adding each layer of composite material separately has now been superseded. ‘Dr Ian Crouch, of the CRCACS, says that by using a technique known as diaphragm forming (DF), the process has been successfully automated. Hawker de Havilland’s manager of business development and strategic planning, Miro Miletic, says although the benefits are small in percentage terms, they are huge in dollar terms’. ‘It’s a process which produces a 2 per cent reduction in terms of raw cost of making composite aircraft components, but overall we are looking at multimillion dollar savings from this process on an annual basis’ Scott (2002, p. 16).

Although there is little published on this micro cluster, it is known that Hawker de Havilland machinery is built by a set of local suppliers (see Royall 2001) and other Australian companies have been successful in winning contracts for aerospace composites, principally in new military aircraft (see Hill 2003). Rather than taking the knowledge back to Seattle, Boeing has been looking to invest in R&D in Australia. Creedy (2002) reported that some of the research for the [discontinued – Masanauskas 2003b] Boeing Sonic Cruiser was to have been done in Australia. The status of that research funding was unclear in mid 2004, but the 7E7 will adopt much of the preliminary research work done for that project. Clearly, this is not a case of outsourced production – where the technology is owned and developed offshore and manufacturing is trans-located, as the technology is locally developed. Because demand is not located in Australia it is reasonable to suggest that this cluster is not an archetypical cluster, yet the involvement of public institutions points to it being a ‘sub-national innovation system’ (spread across two Australian cities) and also points to it being more than purely a corporate GPN. The small group of firms and research groups involved in this cluster are positioning

13

http://www.crc-acs.com.au/

4.4

Clusters, GPNs or Multi-Spatial Cluster Networks?

87

themselves in a global industry with a set of valuable technologies where the entire production system is spatially distributed across the globe. The two dominant players in civil aerospace appear to need to import the Australian expertise in the form of particular products and the Australian cluster has no market without Airbus and Boeing and military aerospace firms. We can now use the case study of the Australian Aerospace micro-cluster to fill in the table proposed earlier. The Australian cluster is involved in both R&D and component manufacturing. The products are sold into the two large centres for aircraft production (USA and Europe) (Table 4.2).

Table 4.2 Linked clustering questions InterIntercompany institutional cluster cluster R&D Design Component design & mfg System mfg System integration Services

4.4.1.4



National Multi-spatial system

Asia connections

North America

Europe











Networked Clustering

Although the aerospace industry is an extreme example of a globally structured value system, on the basis of this analysis, the concept of linked technological and industrial clustering appears to be a valuable way of understanding economic change. This case example does point, however, to the need to determine the production systems to which the relevant concepts are most applicable. This theme is returned to in Chap. 6. Chapter 5 discusses the methodological choices for delving deeper into the spatial structure of interdependencies within a multi-spatial context. In particular, the chapter explores the difficulties of developing a fulsome analysis of linked clustering, in a way that explains the specializations and links of all the nodes in a specific production system. The chapter also explains why it was decided to utilise a multiregional input-output based approach.

Chapter 5

Measuring Inter-Cluster Interdependencies

‘International networking and the resulting exposure of national clusters, however, has a positive impact on innovative activity, reduces costs and increases its economic returns. National systems where firms openly engage in international networking are more successful than closed systems’ (DeBresson et al. 1998, p. 4).

5.1

Identifying Multi-Spatial Innovation Systems

The goal of this project is to contribute to the understanding of the ways separate innovation systems interact within a multi-spatial economic environment. To this point, this has been done through an extensive overarching review of the literature vis-à-vis national innovation systems and sub-national innovation systems. It has also required the need for a specific focus within such a broad research agenda. That focus has been confined here to the role and scale but particularly the spatial structure of cross-border interdependencies. This extends the relationship based perspective that contends that knowledge flows within a local or national environment have fostered clustering and innovation, to those interdependencies that stretch across borders. Ideally, to analyse the latter type of linkages, a researcher would be able to draw upon a wealth of data collected by statistical agencies on the sources of product component being traded across international and regional borders. There would also be ongoing research into the nature of global innovation networks. Unfortunately, the former is hard to determine even for international trade (see Wixted et al. 2006), let alone internationalised innovation (see, e.g. DeBresson et al. 1998). Even within the very specific focus of this book, the lack of previous research forces important choices on the value of a deep study of particular examples (see Giuliani et al. 2005) vs. a study with greater breadth. The latter would lack the case study depth, but nonetheless could outline the general characteristics of the phenomena. It also forces a choice on the degree of emphasis the three foci receive – a choice which is interactive with the choice of depth or breadth. For example, an interest in emphasising the role (and scale in terms of comparative knowledge value) of multi-spatial

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_5, © Springer-Verlag Berlin Heidelberg 2009

89

90

5

Measuring Inter-Cluster Interdependencies

relationships would require a different methodology to one aimed at an interest in the spatial structure (with relevant scaling) of relationships.

5.1.1

The Appeal of I-O Modelling

With little existing research, it was decided that determining a few of the dominant characteristics of multi-spatial innovation systems would be more valuable than an in-depth study that would be lost against the weight of research on proximity. Adopting an approach that could rank the value of extra-territorial linkages for different industries as well as emphasising the spatial structure of relationships was, therefore, deemed the important criteria in choosing the methodology. Research that is broadly similar in focusing on trade relationships has had a tendency to rely upon exports (see Giuliani et al. 2005) and more rarely import data. The desire here was to move away from just indicators of the output of a production system (exports) or the inputs (imports) but to focus attention on the integration and use of traded products. Trade data cannot be easily used to examine the use of the imports. The measurement of trade in the same industry in both directions (intra-industry trade) has been as far as trade analysis can be developed for user-producer connectivity. Traditional trade analysis, therefore, tends to focus simply on the increasing amount of trade, particularly intra-industry trade, and on changes in market share.1 Analysis of changes in market share, however, only bring into sharper relief the need for an assessment of interdependencies because it highlights the way countries often grow their foreign market shares but loose much less across time. Of all the possible methodologies, inter-regional input-output (I-O) modelling offers a way of integrating regional production data (in this case for reasons explained later in this chapter – national data), and trade data to calculate the destination of products and the scale of their importance as a share of industry output. The great insight of Wassily Leontief, the founder of I-O economics (see for example Landefeld and McCulla 1999), was that industries rely upon each other in a system of mutual interdependency. Much of the use of input-output modelling has been for policy purposes. Rose and Miernyk (1989), for example, describe three general economic analyses that were typically of interest, namely; hypothesis testing, policy analysis and economic planning. For advanced market economies, only the first two are of real interest. Hypothesis testing revolves around assessing economic theory against empirical evidence, often on the basis of the factor of production (capital, labour, etc). The current project uses I-O in a more straightforward manner, that of tracing the lines of supply and use across national borders.

1 See Wixted 2005 (appendix 1) on the ratcheting up of export market share. He shows that for a wide cross section of countries and industries for a significant time period, nations did not tend to loose export market share. Such evidence perhaps indicates that Pred’s theory of interdependencies as a key dynamic in the hierarchical stability of American cities is also applicable in international contexts.

5.1

Identifying Multi-Spatial Innovation Systems

5.1.2

91

Criticisms of Using I-O in Innovation Studies

This choice would, however, not be without its critics. There are a few recurring themes in the criticisms of using (I-O) data as a framework for innovation studies. Typically, I-O is dismissed, first because it is based in the production and transfer of goods and not the generation, diffusion and sales of new knowledge and innovations. Second, I-O is based in static equations that are derived from surveys of businesses to determine their use of products from the range of different industries. I-O modelling does not typically incorporate changes to the qualities of the products or dynamic processes such as knowledge accumulation and other feed back effects, so strongly advocated by innovation scholars. By way of example, Carlsson et al. are not alone in their criticisms, continually referring to I-O as a static representation. They say ‘the links among the components of the system are basically one-way, i.e. the system is static’ (2002, p. 235). In contrast to input-output analysis, they wish for data such as that produced by Dahmén, which incorporated disequilibrium and could reveal how the ‘output of the system not only grows over time, but also changes in character and content’ (2002, p. 236). Dahmén’s analysis, however, exists for only one period in time (the interwar years) and only for Sweden. His studies, and others like them, are very difficult to replicate consistently across time and countries. This makes analysis of the character and configuration of innovation systems through time very complex. More important than whether I-O is a static model and therefore inappropriate is determining the types of questions to be addressed. Carlsson et al. have a particular interest in the operations of existing technologies and the development of new technologies. Their description of the networks of people involved in technological systems is not dissimilar to the discussions in Chap. 3 on the drivers of clustering. ‘Technological systems involve market and non-market interaction in three types of network: buyer–supplier (input/output) relationships, problem-solving networks, and informal networks. While there may be considerable overlap between these networks, it is the problem-solving network which really defines both the nature and the boundaries of the system: where do various actors in the system turn for help in solving technical problems? Buyer–supplier linkages are important, the more so the more technical information is transmitted along with the transactions and less so, the more commodity-like the transactions are’ (2002, p. 237).

Particularly for these authors, the interest is in ‘problem solving networks’ and the technical details of innovation and have approximated that interest with the operations of the system as a whole. Yet, there are many dimensions to a given system that can be examined – not just the knowledge flows. What technologies, what role for each type of network, what scale of economic activity and where are these systems located, etc, are all valid questions on the operations of systems of innovation. No single methodology could address all these questions, yet I-O is criticised because it does not provide answers on the specific questions these authors want to answer. A more fruitful approach to analysing available methodologies would seem to be to match the different questions with different analytical approaches, rather than equating an interest in innovation systems with one perspective.

92

5.1.3

5

Measuring Inter-Cluster Interdependencies

Benefits of I-O Modelling

Input-output data is good for what it is designed to do – provide information on the economic structure of relationships within a given spatial territory or extended across territories. Every economy has a unique structure of relations between businesses. Each economy has a different set of industries with different technical input coefficients within different geographic settings. The strength of input-output data is that it reveals these structures clearly and in a way which can be compared across countries and time. Input-output analysis can thus be very useful for overcoming the deficiencies identified by Markusen (1999) who argued that regional and cluster studies often lack information on the wider economic context. I-O data has been gathered for a large number of countries over a long time span so it is possible to investigate the changing structure of relations through time and cross-nationally. Models based on input-output data have been very frequently used for measuring the effects on economic growth arising from changes to government policy, new economic activities or new trading arrangements. Methodologies have been developed for considering the flows of goods between different regions and, as a statistical tool, I-O is amenable to filtering out data noise and focusing on significant information. Another role for input-output economics is its potential use in examining global production networks. While, case studies can highlight the role of imported vehicle components in the auto industry, for example, (see Dicken 2003) these studies tend to underestimate the value added that accumulates to assembler regions because they are focused on imports and production out-sourcing. The purchases of local services, for example, including utilities and the wages for local workers appear to be ignored. Alternatively, as is pointed out by Polenske and Hewings (2004), there is a need to conduct input-output studies of individual production systems. Apart from these general benefits, input-output economics offers very specific advantages to a study of the multi-spatial characteristics of innovation systems. 5.1.3.1

Input-Output and Knowledge and Innovation Interdependencies

Although, as has already been discussed I-O information does not provide some important evidence on knowledge systems, neither is it totally unrelated. When a company markets a product, that product embodies the knowledge that has been used in its creation. The business-to-business component of I-O (intermediate use transactions table) in combination with industry research and expenditure data can be used to analyse the flow of transactions weighted by the embodied R&D. This facilitates an analysis of the spillover of knowledge between industries. Standard R&D intensity measures of industry investment in scientific knowledge do not include the knowledge being accessed through purchasing components or capital equipment. Industries with lower R&D expenditures can rely on knowledge developed in other industries and those with higher R&D expenditures can require a complex range of components.

5.1

Identifying Multi-Spatial Innovation Systems

93

Input-output transactions tables enable a construction of models of embodied knowledge which flows from producers to users. Domestic flows of this sort have been calculated for Germany, Japan and the USA by During and Schnabl (2000) and for Germany, the United Kingdom, Japan and the USA by Drejer (2000). International flows of embodied technology have been examined by Papaconstantinou et al. (1996) and Laursen and Meliciani (2002). The latter authors argue that in general imported technology do not contribute significantly to competitiveness and Papaconstantinou et al. (1996) note that larger economies rely less on imported technology than smaller economies, which they argue can source more than 50% of their technology from foreign markets. Papaconstantinou et al. (1996) reveal that the trend across time has been for increasing reliance on imported technology with, Germany and the USA being the major sources. ICT purchases accounted for the majority of the imported technology. Further, as has already been discussed in Chap. 3 user-producer relationships is one of the few analytical variables that can be used to predict possible sources of innovation. DeBresson’s analyses of business product innovations led him to construct innovation interdependency matrices and to later conclude that these were similar to national intermediate production activity input-output matrices (1991, 1996, 1998 et al., and 1999). The statistical correlation between their matrix of the supply and use of innovative activity and the standard I-O matrix for Italy was 0.836 (DeBresson et al. 1996). Unfortunately, it is not yet established for most countries that national input-output matrices have a high degree of correlation with innovative activity supply and use tables. DeBresson (et al. 1996), however, explain that the high association between innovation transactions and traded transactions: ‘appeal[s] to common sense. They are also consistent with economic theory, which has related the orientation of innovative efforts with demand. …In other words the innovative output of one supplier industry will likely be used in greater proportion by a user industry that consumes more of that industry’s output’ (1996, p. 115).

These findings are only for a domestic setting and might not apply to an extension of such relationships across national borders. But, on the basis of the arguments by DeBresson and the literature on industrial marketing (see for example Thompson et al. 1998) an assumption that the relationships do cross national borders would appear to be consistent with economic theory, the evidence on the social nature of trade and the role of GPNs in transferring knowledge. After all, why should userproducer relationships be co-terminus with borders? Instead, evidence presented by DeBresson et al. (1998) points to international linkages being important for corporate innovativeness. They comment: ‘International networking and the resulting exposure of national clusters, however, has a positive impact on innovative activity, reduces costs and increases its economic returns. National systems where firms openly engage in international networking are more successful than closed systems’ (1998, p. 4).

The ongoing research project upon which this book is based has, therefore, proceeded on the basis that input-output inter-industry linkages, both domestically and internationally, would be significantly correlated to innovative user-producer relations.

94

5

Measuring Inter-Cluster Interdependencies

Ideally, in time, there may be the widespread collection of the type of innovation relationships data that DeBresson has pioneered, but in the meantime a combination of existing input-output analyses for an understanding of the production structure of relationships, case studies of innovative products and cluster developments need to be progressed together. 5.1.3.2

I-O Clusters

Beyond closely allied uses such as those just outlined, input-output analysis has also been used widely in cluster studies (see Chap. 3), despite the criticisms (noted in this chapter) on their use in innovation systems studies. I-O data has been used to identify clusters geographically and industrially (which branches of industry are connected to a cluster and which are not) and to describe the linkages within a given cluster of interest. Remembering again that I-O studies of regional agglomeration date back to Isard and Schooler (1959), there have been recent developments. A major report on clustering in the United Kingdom (Trends Business Research 2001) identifies regional and national clusters using a combination of input-output, employment and value added data at a very detailed level of industry data. Using a combination of national cluster ‘templates’ constructed using national input-output data and location quotient analyses, Feser and Bergman (2000), describe the structure of clusters at the regional level in the USA (U.S. States). Other authors have used input-output methods as a basis for describing particular clusters within a given country. Hauknes (1999) examines the agro-food, oil and gas, construction, transport, paper and graphics and information intensive clusters in Norway. Marceau (1999) has used input-output data to look at the overall structure of manufacturing clusters in Australia whilst den Hertog and Maltha (1999) used very detailed data to investigate the linkages amongst information and communications industries in the Netherlands. Alternatively, both Verbeek (1999) and Hoen (2002a) have developed algorithms for defining clusters on the basis of input-output relations. Hoen’s paper identifies a number of common problems with cluster identification using I-O and notes that standard linkage measures tend to leave large (left over) systems of interconnections – or what he calls mega-clusters. He then proposes a means of calculating relatedness on the basis of significance levels – which appears to be able to robustly identify linked activities (clusters) across different national tables and based either on the Leontief inverse, or the intermediate transactions tables. This later achievement is an advance on other methodologies which produce different clusters depending upon which table is used. However, to date this work on cluster identification has been mostly limited to single region models. Using bilateral trade data in combination with input-output tables that separate domestic use from imported intermediate goods facilitates the multi-spatial inter-regional analysis desired here. Imports can be utilised in industries other than their own and industries in different locations have different mixes of imports from different industries. There is a long tradition of using I-O in this way

5.1

Identifying Multi-Spatial Innovation Systems

95

to examine industrial interdependencies across boundaries which can be traced back to Isard (see Rey 2004, Hewings et al. 2004 and Jackson 2004). Much of this research effort has been in assessing traditional economics questions related to factors of production and the Heckscher–Olin theory of trade (see Polenske and Hewings 2004). There has also been a tradition of measuring the changing structure of linkages between industries and regions. It was noted earlier that economic development at different spatial scales seems to exhibit some similar features such as uneven growth and rigidities which block overall economic convergence. Another feature which seems to characterise different spatial scales within a multi-spatial context is the shifting structure of production and relations. As one example, analysis of the Chicago economy has revealed that between 1980 and 2000 there was a decline in intra-metropolitan transactions in manufacturing industries, but service industries increased their levels of interaction with other sectors. While manufacturing output in Chicago did not decrease, the intensity of intra-metropolitan linkages did decrease and the inter-regional manufacturing interactions increased (Guo et al. 2003). This later evidence confirms earlier material on the hollowing out of the Chicago economy (Sonis and Hewings 2001 and Hewings et al. 1998). These changing patterns of relationships across boundaries can be related to the processes of international production fragmentation (discussed in Chap. 4). One implication of such findings is that regional US economies appear to be concentrating their specialisations and possibly de-specialising their partner relations as they increase their inter-regional linkages – a finding that was made by Laursen (1998b) for OECD bilateral international trade. International I-O modelling supports similar findings for Europe as it does for the mid-west of America. Van der Linden confirms that countries in Europe increased their sectoral specialisation and spatial concentration between 1965 and 1985 with the inter-country patterns ‘hardly changed’ (1998, p. 267). Interdependence grew particularly to 1975, with the small countries most dependent on the larger economies of Europe. Once again, neighbours have been found to have the strongest inter-linkages (so called ‘gravity’ flows). These connections give rise to between 10–15 ‘íntercountry clusters’ (p. 268). However, while these conclusions are in the same direction as the expectations of the Heckscher-Ohlin theory of trade developing towards higher levels of specialisation and greater interdependencies, both van der Linden (1998) and Hoen (2002b) report a level of change for European countries, that is smaller and slower than expected. Further, bilateral trade patterns did not specialise within the EU as also would be expected (Hoen 2002b). These existing bodies of research show the utility of multi-regional input-output techniques, which combines trade information with domestic interactions and other value added items such as wages. The advantages of adopting an inter-regional input-output methodology for this study are thus the ability to: • Assess the whole value added chain, not just trade volumes; • ‘Map’ linkages between countries; • Evaluate the importance of particular source countries in particular sectors and not rely simply on intra-industry trade assumptions; and

96

5

Measuring Inter-Cluster Interdependencies

• Relate the data on value chains to innovation theory through existing embodied knowledge and user-producer relations analyses. Choosing an I-O approach for research on the scale and spatial structure of interdependencies and for better understanding the multi-spatial nature of innovation systems can therefore be justified. However, there are only a few databases that can be utilised in constructing multi-country tables.

5.2

Multi-Country Input-Output Data

Worldwide there are at least six primary harmonized multi-region input-output datasets available (some need to be purchased) and four of them are country level sets. They are: • Bilateral Asian tables covering a number of countries such as Korea and Japan for period 1970 onwards with a very detailed industry classification (produced by the Institute for Developing Economies); • Eurostat tables for 1995 for 14 European Union countries plus a single table covering all 15 members (EuroStat 2000), utilising the NACE-CLIO 25 sector classification, and the dataset continues to expand encompassing more countries and industry categories; • OECD Input-Output Database with good coverage of industries and 9 important economies for years 1970–1990, 20 countries for mid 1990s and then over 30 countries for 2000 (OECD 1995, 2002e, and 2006); • University of Groningen’s European I-O tables; • Statistics Canada’s inter-provincial input-output accounts. Although the Asian tables are excellent, the electronic version of the data has at least in the past been extremely difficult to use and acquiring the data electronically from the paper version is very time consuming. This set of data is also the most costly. The exciting feature of the EU data is the number of countries and the variability in economic size of its coverage (it includes countries like Belgium through to Germany. Results from using this data are presented in Chaps. 6 through 9. There are, however, a number of important disadvantages with this dataset. The first limitation is that due to its focus (European Union) it naturally does not include important economies such as the USA and Japan, the world’s biggest economies. Secondly, just relying upon the EU data would unintentionally raise again the concept of the European supranational innovation system, an idea that has already been critiqued. The third problem with the Eurostat data is the industry classification. As one example, the European data does not disaggregate aircraft industries from other transport, nor is non-ferrous metals separated from iron and steel in the classification. This combines two technology intensive industries with very different industrial structures. Such an aggregation hides too much information. Therefore, the Eurostat classification, even with latter datasets is not ideal.

5.3

Inter-Regional I-O Modelling

97

On the other hand, the OECD data (1995, 2002e and 2006) has been updated and has progressed from initially including relatively few countries but importantly three non-European G8 countries (the USA, Japan and Canada) for the period 1970–1990. It now includes more than 30 countries (see Yamano et al. 2006) which actually generates problems with accessing enough high quality trade data to make use of it all. The OECD data is generally better suited to studies of the type undertaken here in terms of the greater level of industry classification disaggregation and its relative country stability across time periods. It should be noted however, that for various industry categories, EU countries do not provide greater industry detail. The importance of such disaggregation becomes clearer with the presentation of the results in Chap. 6 on the different volume of imported components by different industrial segments, and in Chaps. 7 and 8 on auto and aerospace related cluster complexes. For the analyses presented here, the Eurostat 1995 and OECD time series data have been relied upon. For a full comparison of the OECD and EU 15 model activity category confer with Wixted (2005), while Yamano and Ahmad (2006) provide details of the evolution of the OECD database. Unfortunately, the selection of two inter-country data systems might leave the impression that the concept of multi-spatial innovation systems analysis is understood here as the equivalent of internationalised innovation systems – which is definitely not the intention. If an intra-national inter-regional dataset could have been analysed in a timely manner, it would most likely have been included here. Several alternatives exist for such work in the future. Statistics Canada, for example, has developed a time series of provincial I-O and trade tables and the Regional Research Initiative at West Virginia University has had in the past inter-regional commodity flow data for the USA.

5.3 5.3.1

Inter-Regional I-O Modelling Constructing the I-O Tables

The general data structure2 of the OECD and EU tables are quite straightforward. For each country there are separate tables for ‘domestic transactions’, ‘import transactions’ and ‘total transactions’. The transactions matrix is a table in which the intermediate sales between industries are detailed in a square matrix (see Fig. 5.1). Each of the three tables have this same structure.

2 There are some differences. For example, in the older OECD database (1995) includes tables in constant prices and capital formation). Also, the industry detail between I-O datasets varies. For a description of the EU tables used here see Wixted (2005: 361–364) and for the detailed description of the OECD databases see Yamano and Ahmad (2006).

98

5

Measuring Inter-Cluster Interdependencies

Fig. 5.1 Structure of a domestic Input-output table

In the domestic table underneath the industry-to-industry transactions there are the additional sectors of wages, taxes and value added, etc. that ensure the inputs total 100% of output. To the right of the industry-to-industry matrix the transactions table includes final demand sectors (consumption and exports, etc.) to ensure that all destinations of industry production tally 100%. It is important to note that multi-country datasets, such as those for the EU or OECD, only have information on the use of imports by domestic industry and the likely source industry, at the aggregate level not by supplying countries. The first stage in inter-regional3 modelling is to build a transactions table with a similar structure to that of a single region table. In the figure below each square represents a matrix of all the transactions for a single country seen in the previous diagram. The blacked squares are domestic industry-to-industry transaction matrices (Fig. 5.2). To develop an inter-regional model it is necessary to separate out the various source countries of the imports. This is essential in expanding the square matrix from one country through to being inclusive of all the countries desired in the modelling exercise. It starts with a single imports table for each country (Fig. 5.3). In general, a similar process4 was used for developing the trade transactions matrices for all inter-country models. Each imports spreadsheet was split into separate matrices for (at least) each of the countries in the model plus one other for the rest

3 Inter-regional modelling is distinguished from multi-regional modeling on the basis of whether there is information on all user industries. Multi-regional models by this definition are more limited (see Polenske 1995 and Polenske and Hewings 2004). 4 Other descriptions of the process can be found in Wixted (2005), Wixted et al. (2006) and Wixted and Cooper (2007).

5.3

Inter-Regional I-O Modelling

99

Fig. 5.2 Building up a multi-regional input-output transactions table

Fig. 5.3 Imports transactions tables

of the world. For the OECD model, there is a direct match between industries in the original tables and the OECD Bilateral Trade Database (BTD), with the exception of the service industries. The trade between countries internal to the model and the Rest of the World (ROW) countries was calculated directly from trade flows data in the OECD BTD (2002b). Due to the lack of detailed services trade data, in the modelling of the latest data it is possible to use bilateral trade at the aggregate level (all services) for most OECD countries. For non-OECD countries and for the earlier modelling periods (1970 and 1990) it is necessary to use a modelling assumption of average bilateral

100

5

Measuring Inter-Cluster Interdependencies

manufacturing trade direction. Sensitivity analysis of this method reveals that using manufacturing trade direction is initially okay but is increasingly a poor reflection of services trade direction because of the rise of East Asian manufacturing and the continued strength of the USA and UK in services. It should be noted that services trade data continues improve. The end result of this process can be seen in Fig. 5.4 where the transactions table has been transformed with the inclusion of data in all the regional blocks. At present, it is straightforward to calculate the overall trade preferences (total imports divided by imports from country A) on a row (supply) basis. There is no information that informs us of the purchasing preference of individual industries in a particular country for products from a particular country. Presently there does not appear to be a logical approach to calculating appropriate propensities on a column basis or alternatively a row-column combination to generate a unique ratio for each cell. Presumably, in the absence of actual information some assumptions could be devised on the basis of quality, for example. These are not explored here. For a comparison of the differing scales of the inter-regional models behind the data analysis presented in this book, Fig. 5.5 is provided. In this figure, each cell represents matrices of 33*33 industries (OECD’s 1995 database), 25*25 sectors in the EU database, 32*32 industries (OECD 2002e), and 42*42 industries for 2000 (OECD 2006). Although I-O data for Mexico was not available at the time of writing, it was included in the model as a trade partner (in a similar way to how the Rest of the World was included). The 2000 model contains nearly 1,000,000 cells of data.

Fig. 5.4 A full multi-regional transactions table

5.3

Inter-Regional I-O Modelling

101

Fig. 5.5 Scales of the different Models used to construct the analysis

5.3.2

Limitations of Trade Data

Before going further, it is important to make a few observations about trade data. Increasingly, statistics on national trade performance are becoming problematic. There is a significant deficiency of data on the activities of companies, trade in intermediate products and, as already noted above, for industry level bilateral services trade flows. Intra-firm statistics, for example, are still poor, despite multinational enterprises controlling a substantial proportion of trade.5 While companies are legal entities, ‘industries’ are the creation of statisticians. For example, in the official United Nations definitions for ISIC Rev 3 code 30 is office, accounting and computing machinery6 which includes ‘complete digital systems’ but

5 For countries such as the USA 25% of exports are with affiliated businesses, whilst for others such as Sweden 80% of exports by foreign controlled affiliates were to other affiliates in the same business group – see OECD (2003b) pp108–109. 6 office, accounting and computing machinery http://unstats.un.org/unsd/cr/registry/regcs.asp?Cl = 2&Lg = 1&Co = 30

102

5

Measuring Inter-Cluster Interdependencies

excludes computing components such as semiconductors which are in a completely separate division (32).7 Therefore, sophisticated electronic games (and associated software) could be split across 3,210 (the components), 3,000 (complete units), 3,694 (manufacture of games and toys [the electronic equipment]), 7,220 (the writing of the software), and 2,230 (reproduction of recorded media (the mass production of the software on disks). Reading the fine print of classifications is tedious but essential. Thus, the manipulation of the data engaged in here only approximates reality and it should not be taken as more than a gross approximation of the changing realities of international business. Finally, trade data is traditionally aggregated to the national level. Therefore, although we are aware that the regional agglomeration of industry is an important characteristic of economies (noted in Chaps. 3 and 4), there is little data on how such clusters or regions trade either domestically or internationally. Nevertheless, such approximations are useful and the analysis of spatial structures valuable in opening new research areas and beginning to flesh out the realities of ‘production globalisation’.

5.3.3

Inter-Country Modelling Analyses

In contrast to other analyses of inter-country inter-industry linkages which use the ‘hypothetical extraction’ methodology or other production output based methodologies, the current research project adopts the net multiplier techniques developed by Cooper (2000). Dietzenbacher and van der Linden (1997), van der Linden (1998) and Hoen (2002b) employ a modified Strassert method that enables the “hypothetical extraction” of industries to identify backward and forward linkages across EU countries. As is usual with this method, the results are reported in terms of output effects. The hypothetical extraction technique proceeds by progressively removing sectors and then modelling the effects of that sector on the rest of the system. This works as if each sector were, in turn, considered as only importing with no domestic activity. It is a good methodology for understanding so-called ‘key sectors’ – those that have a greater influence on production and productivity than others. In this way, it also identifies the linkages to other sectors and regions. Using a third technique, van der Linden et al. (2000) identify propulsive, reactive and dependent sectors. One of the important points they make is that despite the apparent integration of the European Union, inter-country value added spillovers are quite small. Although this ‘fields of influence’ approach is quite a different technique to the hypothetical extraction method, it also concentrates on measuring production effects. An additional common feature is that it can be used for understanding both backward and forward linkages. Van der Linden et al. (2000) use the methodology for tracing the effects of technological change (the technical coefficients) through linkages rather than measuring the strength of the linkages. More 7 3210 Manufacture of electronic valves and tubes and other electronic components within Division 32 Manufacture of radio, television and communication equipment and apparatus.

5.3

Inter-Regional I-O Modelling

103

recently, Dietzenbacher and Romero (2007) have used the concept of propagation length to examine inter-country clusters. The developed approach of ‘average propagation length’ (APL) is still based on the hypothetical extraction methodology, but leads to interesting results on the flows of goods and services between countries. This approach has the advantage of breaking down separate stages of production, an analytical outcome that can’t be achieved with the methodology adopted here. Both hypothetical extraction and fields of influence are capable of identifying trans-border clusters. The hypothetical extraction method is able to indicate which sectors and countries will be most affected by the lack of a sector that would otherwise provide a market for their intermediate goods and services. Nevertheless, because of the sheer size of production effects, the number of sectors in a multi-sector and multi-country model that one would need to sequentially extract and the smallness of inter-country spillovers relative to domestic output effects, there is a case for considering other approaches in terms of both the methodology and the effects that are measured. Similar points apply to the fields of influence approach. In addition, there is a case to examine the existence of trans-border clusters quite independently of what might be the case under alternative technology scenarios. One objective in modelling the interactions of economies is to search for ‘above normal’ inter-country spillovers, which are ‘normalised’ so that the small intercountry spillovers are not dominated by the naturally larger intra-country effects, so the net multiplier approach was seen as particularly useful.

5.3.4

Net Multipliers

The methodology used here for measuring the networking of clusters in the European Union and the OECD is therefore different to the application of input-output analysis in current inter-country EU models. The focus here is on tracing flows of output arising indirectly through inter-industry and inter-country linkages via a backward linkages approach. Essentially, we need to obtain partial sums of backward indirect production multipliers using output coefficients as insights. The measures are based on indirect output flows to examine the strengths of linkages between a sector and a region (not necessarily the home region of the sector). One measure of the importance of a sector to a region is its contribution to indirect output. Such a ‘net multipliers’ approach facilitates a focus on the multi-regional linkages and has specific advantages over the existing methodologies proposed in the literature on input-output economics. By way of example, Dietzenbacher and van der Linden highlight the results of their modelling of the industrial interactions of European economies. ‘For example, hypothetically removing the German food sector yields the following reductions in the outputs (all as a percentage of this sector’s actual output) 82.0 in Germany, 14.2 in the other six EC countries, and 96.2 in total (see Table 1). Therefore 15% of the total backward linkages is intercountry and 85% is domestic’ (1997, p. 244).

104

5

Measuring Inter-Cluster Interdependencies

Because the hypothetical extraction methodology is deigned to provide output effects, the numbers as seen in the quote, are in relation to output and are quite difficult to relate to the literature on trade and the internationalisation of production. By contrast, Cooper’s methodology for calculating net output applied here gives a much more interpretable result. An increase in the domestic demand for food in Germany in 1995 would result in the following division of activity: • 53.72% of the increase would be direct German effects; • 27.44% would be indirect value added, remaining in Germany; • Imports from the other 14 economies of the EU would account for 9.2% of the demand increase (indirect); and • Imports from the rest of the world would account for the remaining 9.7% of the output increase (indirect). This latter presentation of modelling results is attractive because it focuses attention on easily interpretable indicators of production internationalisation and bilateral linkages. In this way, it can be directly related to the extant body of literature on globalization and trade. It also lends itself to be illustrated with network diagrams. Since the results required from this modelling project were non-standard for input-output modelling and because the scale of the models is large (see Fig. 5.5), a new software8 application that utilises the Microsoft Visual Basic programming language in Excel has been used. The resulting software is quite flexible as it allows an analyst to select which regions will be modelled, but automates the location of relevant regional names within the dataset. The software will generate a Leontief inverse, calculate the coefficients matrices and also create a set of output results matrices which match in layout (cell for cell) the original transactions table. The software calculates these output results by increasing demand by 1 unit in all industries in all selected countries. Whilst the software generates the linkages effects through a series of simultaneous equations calculating the value of a linkage from the total effect on an economy and it is designed to ensure that the output distributed across industries and countries still equals the additional demand (1 unit). The current version of the software does not use the final demand vectors of the input-output tables nor does it permit variable increases in industry demand. The algebraic expression of the modelling has been previously published and for this reason has not been replicated here. Interested readers are directed to Cooper (2000) for the early development, Wixted (2005) and Wixted and Cooper (2007). The latter publication as a related work to the current book is based on the same methodology and so the algebra is identical.9 8 Spatial Connect was primarily developed by Professor Russel Cooper, with the author of this book focused on generating the inter-regional transactions tables. 9 It should be noted here that although Wixted and Cooper (2007) created a dataset for 2000 it was created before the OECD released their I-O data for the year 2000 and was based on some basic forecasting assumptions.

5.4

5.4

Data Analysis: Chaps. 6–9

105

Data Analysis: Chaps. 6–9

The production of linkage values for every interaction of user and supplier industries across as many as 22 economies plus exogenous regions in the latest OECD data generates enormous databases. It is possible to conceive of a large number of tests that could be developed to run on this dataset. Two primary analyses have been chosen to provide a structured overview of the results to address the central themes of the scale and spatial structure of trans-border connections whilst leaving deeper analysis to future work that can focus on generating detailed data for case studies, namely: • The broad patterns of the internationalisation of production, including analysis of product complexity (Chap. 6); and • Mapping the spatial architecture of production for the motor vehicle (Chap. 7), aerospace (Chap. 8) and electronics and ICT (Chap. 9) production systems. The starting point for the presentation of the results is to summarise them in terms of which industries require the most imports per dollar or European Currency Unit (ECU) of output (Chap. 6). This analysis indicates which industries have highly internationalised production systems and interestingly motor vehicles, aerospace, ICT and electronics standout. The data collected from the modelling is then supplemented by an analysis of the industry characteristics that may account for why these higher R&D intensive industries require a significant level of imports. Going the next step, the data output of the modelling can be aggregated into tables of inputs into industries in the different regions. This reveals how much of a particular industry’s total output is imported from each trading partner. The data can then be processed to reveal whether the strength of particular linkages is significant or weak. There is no ‘perfect’ statistical level of significance to determine strong and weak ties in these cases. As the level is lowered, more links are revealed and the diagrams become more challenging to interpret. If the measure is too high, the diagrams become increasingly bare and stylistic (see Wixted 2005 for comparative charts based on some of the same data). It was decided that a cut-off of 10% of imports (note not of total output) optimized this balance. This benchmark combines the visual benefit with a measurement advantage. Until the beginnings of the 1990s, it could be argued that there were roughly 20–25 significant OECD industrialized type economies in the world.10 Fixing the backwards and forward measure of significance at the same point of 10% requires a threshold of at least twice an ‘average’ share of the supply of goods for the years before the 1990s and an increasingly important slice beyond this point. With economies, such as South Korea, and more recently China establishing important global trading positions, this indicator will assist in focusing attention on both emerging and stable trade relationships.

10

OECD membership during this period grew to 24 countries.

106

5

Measuring Inter-Cluster Interdependencies

This focus on creating visualisations based on share of imports also minimises as far as possible the negative effects of the scale factor in this analysis. As was noted earlier, and it should be no surprise – as the scale of what is considered to be a single economy (e.g. the USA) increases, the importance of imports declines. If the statistical measure of external linkages to be used was based on total output then the larger economies would show no external interdependencies. Using a share of imports reveals the connections that are important for each economy. One important note is that Chaps. 7 (motor vehicle production), 8 (civil aerospace) and 9 (ICT) introduce the idea of cluster ‘hierarchy’. This word is used purely in an economic sense, as does Rose and Miernyk, going back to Leontief, who indicates that there are: ‘four major concepts of structural analysis: interdependence, dependence, hierarchy and circularity. The first two refer to the extent to which an economy is composed of enclaves or interrelated industries. Hierarchy refers to the economic pattern of primary, secondary and tertiary production and their detailed components. Circularity, or robustness, refers to the extent of intermediate good requirements for production’ (1989, p. 245).

Hierarchy is not intended to suggest centres of control, political power, leadership, imperial power or other sociological interpretations, even though other authors may wish to pursue such lines of inquiry. Here, it simply refers to the degree to which a particular cluster is important in regard to other clusters.

Chapter 6

Clustering Internationalisation

‘Finding the source of demand that may spark off the growth of the cluster can be critical for its rise, and in many respects it should be one of the policy focuses of this area’ (Bresnahan et al. 2001, p. 843).

6.1

Dependence on Imported Components

It has been shown already that the systems of innovation research agenda at whatever scale of cluster or nations, has been defined, predominantly, within national borders. The research has emphasised the endogenous and proximity based characteristics of systemic innovation. Chapter 5 suggested that the structure of multispatial systems of innovation could be investigated though inter-country input-output modelling. The next four chapters present a series of different perspectives on the scale, and spatial structure of inter-regional interdependencies. This is achieved through specifically examining the specialisation of locations, globalised products and the strong trade linkages between nations. In this chapter, the data on imports is compared at the level of the aggregate value of transfers across borders arising from demand in each industry in each country. As such, this chapter is directed at determining one element of the scale of inter-regional linkages and to highlight one role of imports (embodied technology). The modelling results can be used to measure the degree to which each industry in each country relies on international inputs (imports) per unit of output. The results of the modelling are analysed to seek commonalities between the EU and OECD models, both for their significance to internationalisation debates and as a way of checking the validity of the models. One of the outcomes of this analysis is that a number of technology-based industries have high import requirements. This indicates they have highly internationalised supply chains. As these industries are typically perceived as knowledge-based and heavily dependent on national or local systems of innovation, this is an important result. Analysis presented here begins to examine what is known about these industries with the aim of identifying the key characteristics that may account for this need for imports.

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_6, © Springer-Verlag Berlin Heidelberg 2009

107

108

6

Clustering Internationalisation

Inter-country modelling calculates the spread of output requirements across all industries and countries that arises from an increase in demand in a particular country’s industry. In this way, it focuses attention on the industrial ingredients required to boost production. As an example, an increase in the demand for motor vehicles in country A will result in an increase in purchases of rubber, plastics, glass, metals of various sorts and electronic components, etc. in countries A, B, C…. Because the approach adopted here has been to group all the value spillovers that stem from all industries arising from demand increases in each of the database classifications, into national agglomerations, the inter-industry linkages are considered here as national-meso clusters (see Table 3.3). In this chapter and those that follow, the structural patterns of the required ingredients is what matters. Therefore, different clusters are compared on the basis of their differing input requirements and source countries rather than on the dollar or Euro value of their trade per se. The analysis is also restricted to production activities and does not include any investigation of trading patterns in finished goods (final demand). Figure 6.1 provides a diagrammatic representation of the way the import values stemming from industry growth is represented here and in later chapters.

Transport equipment

France Germany’s imports from France in the transport cluster is calculated by summing all supplies to Germany initiated by demand growth in the transport industry. Arrows in this analysis point in the direction of the value added –goods flow in the reverse direction.

Germany

A

B

C

D

E

Fig. 6.1 Domestic and imported content: cluster calculations

Table 6.1 presents the results of summarising the data to the extent of tabularising the top five national-meso clusters by the level of imported content for each of the periods modelled from the available OECD data.

1 1 1

2

2

5 4 2 1 1 1

Textiles, clothing and footwear Industrial chemicals Other transport equip Paper, paper products and printing Wood products and furniture Rubber and plastic products Other manufacturing Mining and quarrying Scientific instruments, etc Food, beverages and tobacco Transport and storage Pharmaceuticals Electrical Machinery Computer services Water Transport Manufacture & distribution of gas through mains Air transport Chemical (excl. Pharmaceuticals)

5 2 3

1

5 (of 8 series) 4 6

3 (of 8 series) 4 6 4

Shipbuilding and repairing ships

8 7 (of 14 series)

9 2 (of 14 series)

Petroleum and coal products Office equipment and Radio, TV and communications equip. Motor vehicles Aircraft Non-ferrous metals Iron and steel

1990 (9)

1970 (9)

Top Importing Clusters

1

1

2

2

3

5 5 4 3

8 11 (of 17 series)

1995 (9)

Table 6.1 Top Importing Clusters of OECD and other major Economies 1970–2000

3 6 2

5

1

1

1

2

5

(continued)

12 4 (of 28 series) 3 7

17 29 (of 43 series)

2000 (22)

7 5

2 2 1

1

2 4

2 2 2

3

12 9 9 8

14 27 (of 38 series)

1995 (20)

3 2

1

6 2 (of five series) 2 (of five series) 3 (incls. non-ferrous in 3 cases) 4 (incls. Aircraft in 4 cases) 1

8 11 (of 17 series)

2000 (9)

6.1 Dependence on Imported Components 109

45 (9 * 5)

Supporting and auxiliary transport activities Community, social & personal services Total 45 (9 * 5)

1990 (9)

45 (9 * 5)

1995 (9)

45 (9 * 5)

2000 (9)

100 (20*5)

1995 (20)

110 (22 * 5)

1

1

2000 (22)

Note 1: In some cases, countries amalgamate data series for a variety of reasons. So, for example, most European countries currently do not report data on the Aircraft and aerospace industry separately. Note 2: The inter-regional modelling for each of the OECD (2006, 2002 and 1995) databases were developed and analysed on a consistent basis. By keeping the original nine countries (Australia, Canada, Denmark, France, Germany, the Netherlands, the UK and the USA) constant it is possible to provide a time series. Although the original OECD (1995) tables included Italy it was only included for one year (mid 1980s) and was therefore excluded from the modelling. As Chap. 5 stated the modelling of the most recent OECD (2006) database has not included all possible countries because of the complexities in generating trade data. Note 3: For comparative purposes, the results of all countries are presented in the shaded tones. Sources: Previous versions of this table have appeared in Wixted (2005), Wixted et al. (2006) and Wixted and Cooper (2007).

1970 (9)

Top Importing Clusters

Table 6.1 (continued)

110 6 Clustering Internationalisation

6.1

Dependence on Imported Components

111

It quickly becomes apparent from this table that products such as processed petroleum and coal, etc., are a basic requirement of modern economies. Thus, in the OECD 1990, 1995 and 2000 models, only one country did not have the petroleum and coal cluster in its list of top five importers: the United Kingdom. This is logical as the UK has access to North Sea oil and thus an important national source. As a contrast, shipbuilding only appeared once in 1990 (Canada). Non-ferrous metals, is surprisingly high on the list, as another resource based commodity. This probably reflects the internationally staged production character of the industry which requires access to cheap energy. However, beyond this, the results become quite interesting both for what is high and low on the list. Remembering that this analysis reveals the degree to which industries require inputs or components from other countries, it has little to say regarding the shifting of nearly complete supply chains (textiles, perhaps as an example, see OECD 2004b). What is striking is that the relatively technology intensive industries of office equip, radio and communications equip, motor vehicles and aerospace appear quite high in the list. These findings are generally supported by the OECD (2007b) using an import penetration measure. Table 6.2 reveals that modelling of 1995 Eurostat data (EU 15) presents some similar results, allowing for a different classification system. In the EU model, transport equipment appears in the list for 12 out of 15 countries. The countries in which the transport equipment cluster does not have a high degree of dependency on imports, relative to other clusters (Germany, Greece and Ireland), is an interesting selection of countries. These three can be understood to a limited extent. As Germany has an economically powerful production system it has little dependency on imports. For Ireland, transport equipment ranks below five other clusters with very high import values and

Table 6.2 EU national-meso clusters: prevalence in the top five importers National meso clusters

#/15 No. in top 5/out of 15 possible instances

Transport equipment Textiles and clothing, leather and footwear Rubber and plastic products Ferrous and non-ferrous ores and metals Chemical products Office and data processing machines Maritime and air transport services Fuel and power products Electrical goods Recovery, repair services, wholesale, retail Services of credit and insurance institutions Metal products except machinery Agricultural and industrial machinery Paper and printing products Total (15 countries*5 = 75)

12 10 10 9 9 6 5 4 3 2 2 1 1 1 75

112

6

Clustering Internationalisation

Greece’s transport equipment industry probably because it has a small underdeveloped economy. Concentrating on those industries that have a high re-occurrence in the top five cluster import intensities for each model, it is possible to compare results. Table 6.3 selects those clusters which occur in 60% of possibilities (5 or more cases in the OECD 1990 model (selected because the base years for the EU data are frequently prior to 1995) and 9 or more for the EU model). Three clusters are common to both the OECD model and the EU model, with one OECD based classification (non-ferrous metals) included as a subset of the wider EU NACE-CLIO1 category of ferrous and non-ferrous ores and metals. As Table 6.3 provides an approximate sensitivity analysis for the two models, it is reassuring that, the two models have produced somewhat related results. In both models, a relatively limited number of industries appear to have the highest requirement for imported intermediate goods. The level of statistical aggregation matters here, as different segments of a particular industry may be more or less reliant on imports. For example, the petroleum industry is well known to be dependent on resources from the Middle East, Africa and the North Sea. However, in the EU data, the classification is fuel and power, which is a much more national activity, because it includes domestic power generation, as well as petroleum. This combination has a low value added spillover (imports) score. Curiously, for 10 of the 15 European countries the import propensity for fuels and power are rated within the band of 12–17th place of the 25 NACE CLIO industry classifications, perhaps suggesting some similarity for the co-efficient of petroleum imports in the overall system structure.

Table 6.3 Similarities of industries in the top 5 in the OECD 1990 and EU 1995 Industry OECD model OECD model OECD model OECD model OECD model EU model EU model EU model EU model EU model EU model

1

Petroleum & coal products Non-ferrous metals Textiles, apparel & leather Motor vehicles Office machines, Transport Textiles & clothing, leather & footwear Rubber and plastic products Chemical products Ferrous and non-ferrous ores and metals Office and data machines

Classification for input-output tables.

Occurrences

Matches

8/9 6/9 5/9 5/9 4/7 12/15 10/15

-ü ü ü ü ü

10/15 9/15 9/15

--

6/15

ü

6.2

Complexity, Clustering and Imports

6.2 6.2.1

113

Complexity, Clustering and Imports Knowledge Economy and Clustering

It has been argued by some, notably Audretsch and Feldman that knowledge and production activities cluster when new knowledge and knowledge spillovers are important. ‘Indeed, we find that a key determinant of the extent to which the location of production is geographically concentrated is the relative importance of new economic knowledge in the industry. Even after controlling for the concentration of production we find evidence that industries in which knowledge spillovers are more prevalent – that is where industry R&D, university research and skilled labor are the most important – have a greater propensity for innovative activity to cluster than industries where knowledge externalities are less important’ (1996, p. 639).

High Low

INTERNATIONALISATION

If industries that rely on new knowledge and which thus have relatively greater levels of research and development (R&D) expenditure are more likely to cluster, they might be expected to import less if, as seems likely, clusters have been assumed synonymous with value chains. The relationship between technology and imports might then look something like that depicted in Fig. 6.2.

Low

High

R&D INTENSITY Fig. 6.2 Conceptual relationship between technology and production imports

Instead, Tables 6.1 and 6.2 have already revealed that a number of high technology national-meso clusters can have high requirements for imports. This highlights an important disjuncture between the research on geographic space and high technology industries and the evidence generated from assessing the need for imports. Nevertheless, expected industries (e.g. textiles, etc.) that are driven by cost-based competitiveness, and which import from low wage countries, did appear on the lists in Tables 6.1 and 6.2. Importantly, however, while the production fragmentation

114

6

Clustering Internationalisation

literature (see Chap. 4) is most interested in the movement of production to lower cost production centres, much of the reported trade in the models presented here move between high-income economies. At the very least, the tabulations suggest the existence of three high import intensity categories. These appear to be based around production costs (TCF), resources and resource processing (petroleum and non-ferrous metals) and those which are relatively technology intensive (motor vehicles, computers and aircraft). To process the data on imports, two existing taxonomies of industry and innovation activity have been utilised in the first instance. These approaches are R&D intensity (the level of R&D expenditure as a percentage of industries value added) and Pavitt’s (1984) taxonomy of the innovation characteristics of industry. Stemming from the results gained from these analyses, it became necessary to develop the outline of a technological complexity (technological breadth v technological depth) based approach to classifying industries.

6.2.2

R&D Intensity

The R&D intensity measure of technology has been widely used as an indicator of innovativeness across more than two decades (see Sandven and Smith 1998 and OECD 2003b). The collection of data on research and development expenditure at the industry level has facilitated the development of a classification of industry based on the intensity of their scientific knowledge inputs. The standard groups are high, medium and low technology, although the medium technology group is now commonly broken into high and low. The advantage of this approach is that it is empirically driven, in that it is based objectively on the average OECD R&D intensity for each industry, it can be calculated for most OECD countries and it can be updated regularly. As Fig. 6.3 reveals, the international averages for industry R&D in the OECD, do separate into broadly dissimilar agglomerations. Low technology R&D intensities are quite tightly packed in the bottom left hand corner of the chart, while the high technology group averages are quite dispersed across the top right quadrant. There are, however, problems with the R&D intensity categories devised by the OECD. Typically, such the categories are adopted as if all countries had similar patterns of R&D intensity. This is not the case as can been seen in the OECD’s own publications (OECD 2001a, Annex 1.2) or in the analysis of Wixted (2005, pp. 174–175). At the country level, there are wide discrepancies between the OECD average and individual country performance. Despite these limitations, R&D intensity is a useful metric to begin to exploring patterns of production internationalised. Table 6.4 breaks down the top five internationalised clusters for each country by their R&D intensity classification, using 1990 as a benchmark year. In Chap. 5 it was explained why the EU industry categories are not as helpful for analytical purposes as those developed by the OECD and for this reason, here slightly more emphasis should be placed on the outcome from the OECD model than the outcome from the EU model.

6.2

Complexity, Clustering and Imports

115

30 Aircraft

25 Pharmaceuticals

R & D as % V alue Added

20 HIGH TECHNOLOGY

Communciations equip

Computing

15

Motor vehicles

Precision instruments

10 Railroad equip Chemicals 5

Electrical mach Total mfg MEDIUM TECHNOLOGY Machinery equip

Petroleum etc

0 0

1

2

3

4

5

6

7

8

9

10

11

R&D as % Production

Fig. 6.3 R&D intensities by industry. Source: Data sourced from OECD (2005: 183)

Within Table 6.4, comparisons of the actual numbers can only be made inside a column, but the broad profile can be compared across the columns. Curiously, while there are differences between the EU model and the OECD model, there is a sense in which the results even out. Leaving aside services, the high technology and low technology categories are both at the bottom end of the distribution, high technology having the lowest in the EU model and low technology the fewest representatives in the OECD model. In the EU medium-high and medium-low have very similar results, whilst medium-high is the same as high tech in the OECD. Clusters in the medium-low category account for highest frequency of highest value (as a share of production) importers in both models. As it turns out, using an R&D intensity based classification does not provide too much illumination. Neither low nor high technology industries appear to be any

116

6

Clustering Internationalisation

Table 6.4 Top importing clusters by technology classification Classification

EU (#/75)

OECD (#/45)

High technology Medium-high technology Medium-low technology Low technology Services

9 22 24 11 9

11 11 16 7 0

Note: As five clusters are selected from each country, there are 75 counts for the EU 15 and 45 counts for the OECD.

more represented than the other. Two industries in the high technology group, instruments and pharmaceuticals, are not intensively internationalised at all, while the other two, aircraft and computers do make it to the list in a number of countries. At the other end of the spectrum in the low technology industries, food and beverages retains a strong local focus, but textiles, clothing and footwear is intensely internationalised. The medium-high technology industry of industrial machinery (machinery n.e.c.) does not make it to the list of high importers, while the mediumlow group of industries, with many resource based processing industries, has the highest score of high import clusters for both models. Returning to examine the proposition that was developed from the comment by Audretsch and Feldman (1996) that new knowledge generation and clustering go together, we could test for a correlation between R&D intensities and the imported output scores. Figure 6.4 is a scatter chart plotting R&D intensity (X axis) against

0.60

0.50

InterVA

0.40

0.30

0.20

0.10

0.00 0.10

1.00

10.00

100.00

R&D

Fig. 6.4 Scatter diagram: R&D intensity and imported value added scatter Source: R&D data (X axis) from OECD 2001a, imports contribution to output (Y axis) from the OECD 1990 model.

6.2

Complexity, Clustering and Imports

117

the percentage of output being imported (Y axis), generated from the OECD 1990 model, for Canada, Denmark, France, Germany, Japan, UK and the USA, for manufacturing based clusters. The scatter apparent in Fig. 6.4 appears to be random. This is in contrast to the original speculation, represented in Fig. 6.2, that high technology industries, based in new knowledge formation may have relied less on imports than lower technology industries. Instead, all variations of possibilities seem to be visible. At this aggregated level of industry data, the broad explanatory power of R&D intensity for the degree to which value is captured nationally thus appears to be poor.

6.2.3

Pavitt’s Industry and Innovation Taxonomy

In the absence of any strong association between R&D expenditures, the imported share of output, and local systems of innovation, it necessary to examine the data using different analytical tools. There have been a number of attempts to develop taxonomies of manufacturing industry that are an alternative to the one based on R&D intensities, but none of them have been widely adopted. One of the more useful approaches was put forward by Pavitt (1984). Pavitt’s taxonomy (1984), while being widely referenced has suffered from the degree of difficulty in updating it, or applying it to countries other than the UK – which is inherent in its original methodology and from a dearth of further development.2 The taxonomy was based on an analysis of a database compiled by the Science Policy Research Unit of business innovations. Pavitt’s examination of innovations enabled him to categorise the innovations on the basis of their characteristics: explaining ‘similarities and differences amongst sectors in the sources, nature and impact of innovations, defined by the sources of knowledge inputs, by the size and principal lines of activity of innovating firms, and by the sectors of innovations’ production and main use’ (1984, p. 343). From this analysis, he suggested four sectoral innovation types: • • • •

Supplier-dominated (agriculture & traditional manufactures such as textiles); Scale-intensive (bulk materials [metals, etc], autos, etc); Specialised-suppliers (machinery and instruments); and Science-based (electronics, electrical and chemicals).

Later the OECD (1996b) modified this taxonomy in a way that has some benefits because it separates out two groups of process-based industries, namely: • Resource-intensive (food, beverages & tobacco, wood products, petroleum refining, non-metallic minerals, and non-ferrous metals); • Labour-intensive (TCF, fabricated metal products, and other manufactures); • Specialised-supplier (non-electrical machinery, electrical machinery, communications and semi-conductors); 2

With notable exceptions DeBresson et al. (1996) and Marsilli and Verspagen 2002

118

6

Clustering Internationalisation

• Scale-intensive (paper & printing, chemicals, rubber & plastics, iron & steel, shipbuilding, motor vehicles, and other transport); and • Science-based (aerospace; computers; pharmaceuticals and instruments). Although the OECD taxonomy revision remains unsatisfactory because the methodology and the data for updating it are obscure. Further, non-ferrous metals processing plants are not necessarily co-located with the resource base and thus this should therefore be treated as a special case and arbitrarily moved from the resource-intensive category back to scale intensive (where Pavitt had originally located it). Although there are no perfect taxonomies of industry – the design and usefulness depends upon on the purpose, Table 6.5 (which repeats the procedure used for 6.4) and indicates that the modified Pavitt is quite useful in identifying some commonalities of high import clusters. In contrast to categories based on R&D intensity, the profile generated from the modified Pavitt taxonomy is much more persuasive. On this analysis, scale intensiveness is clearly an important variable in considering the internationalisation of production. Scale accounting for nearly four times more cluster counts than the next nearest classification for EU activities, while for the OECD it accounts for twice as many. This has considerable logic to it – production configured on the basis of large volumes of merchandise may need quantities of inputs that cannot be all supplied locally. A few scale-based sectors are based on resource processing with minerals likely to account for a large percentage of the imports and which could be sourced from either a wide range of developing economies or high income resource-based economies such as Australia and Canada. While Pavitt’s taxonomy was based in a combination of assessing the knowledge inputs and product users, and the derivation of the OECD taxonomy is unclear, it is possible that these taxonomies should be considered to focus on primary factors, while secondary characteristics may also exist. For example, aerospace is defined as science-based but it may rely also on the scale of operations. Scale in turn could be seen to have the two dimensions of capital and R&D. Figure 6.5 was designed to explore for a connection between capital intensity (a measure of scale intensiveness) and R&D intensity (science base) for six large OECD countries. The top right quadrant of Fig. 6.5 groups industries that are high in knowledge and capital requirements.

Table 6.5 Application of modified Pavitt taxonomy to cluster imports Science based Specialised supplier Scale intensivea Labour intensive Resource intensivea

EU model #/75

OECD model #/45

6 4 41 11 4

8 3 20 5 9

Note: Based on OECD definitions but aresource intensive was modified by shifting non-ferrous metals to the scale intensive group.

0

2

4

6

8

10

12

14

0

1

2

3

4

5

6

8

9

Capital Intensity

7

Machinery nec

Railroad

10

11

Chemicals

12

Motor vehicles

13

14

15

Petroleum prods

Comms equip

16

R&D intense and capital intense industry grouping

Precision instruments

Computing

Pharmaceuticals

Electrical mach

Aircraft

Fig. 6.5 R&D and capital intensity Data source: OECD STAN Database 2002c and 2003c. Calculations based upon Canada., France, Germany, UK and the USA. Capital intensity was calculated for the period 1996–2000 as a share of VA

R&D Intensity

6.2 Complexity, Clustering and Imports 119

120

6

Clustering Internationalisation

A few of the national-meso clusters that appear in the top right quadrant of Fig. 6.5, such as motor vehicles, office and computing and radio and communications have high import requirements, yet the pharmaceuticals and instruments clusters do not. In the bottom right quadrant are a number of industries, which in the modified Pavitt taxonomy appear as either scale based or resource based. These results suggest that capital intensity is a reasonable starting proxy for a data driven approach to the notion of scale intensity. Analysis of Fig. 6.5 reveals that capital intensity and R&D intensity together do isolate a number of the technology-based high import industries. Aerospace, is one exception which has a high import intensity but surprisingly its capital intensity is not very high. Pharmaceuticals and instruments on the other hand both have high capital and R&D requirements but do not have a high import intensity. Intuitively, what separates aerospace from pharmaceuticals is the complex integration of components vs. a simpler but integrated manufacturing process which can be located in any number of countries. Is it then possible to consider the degree of the complexity in the use of intermediate goods?

6.2.3

Complex Technologies

An important dimension in the discussion on technological complexity focuses on the geographic fragmentation of product design and development from manufacturing. Pavitt (2003a, b) has argued that there is an ongoing process of geographically decoupling product design and manufacturing. Such a trend, he argued, would result in the separation of corporations that build components and companies that are ‘system integrators’. Pavitt proposes that the resulting multi-organisational structure of production will reinforce an observed trend towards the offshoring of factories to developing countries. The possibility of the fragmentation of production splintering amongst industrialised countries, rather than to the developing world, each with their own knowledge bases and specialisations was not part of Pavitt’s analysis. Pavitt’s argument of decoupling R&D from production is a direct continuation of the arguments of Patel and Pavitt’s earlier arguments (1991) that technology development was ‘non-globalised’ because it was strongly related to home nation technological specialisations. Curiously, the concept that clusters in advanced economies could be technologically specialised and fit within networks of nodes is consistent with the evidence that countries and clusters maintain their specialisations across time. The geographic distribution of knowledge and production is one dimension. The breadth and depth of knowledge formation in technological complexity, markets, products, production processes and administration and management (p. 813) is another dimension (see Wang & Tunzelmann (2000). Some products require a wide spread of capabilities, in contrast to others that require deep knowledge of just a few scientific fields. The concept of Wang & Tunzelmann (2000) is a good one and one that is relevant to the problem of production fragmentation. There is, however, a wide gap between the fledgling concept and the availability of data.

6.2

Complexity, Clustering and Imports

121

Ozman (2007) in her recent analysis of patenting in biotechnology and telecommunications has attempted to map patenting breadth and depth by looking at the characteristics of the patents themselves. In a similar way for the present project, a unique data set created and published by Patel and Pavitt (1997) on the patenting structures of large corporations in a range of industries was re-configured here. Patel and Pavitt’s approach was to devise an indicator system to designate whether a corporation’s patenting competency in a patent technology class was marginal (1 point), background (2 points), niche (3 points) or core (4 points). They also counted the number of patent fields in which a company was active. The data in their 1997, paper has been reconfigured here (Fig. 6.6) to analyse technological breadth and depth, a purpose that was not in there original purpose but which their construction of the data is perfectly suited. The specialisation scores were summed for a count of technological activity (breadth), whilst the count of core and niche strengths was tallied to indicate the degree of technological depth. Thus: • The X-axis measures the total count of all patenting competency scores allocated by Patel and Pavitt for each industry. An industry with one of each type of competency (1, 2, 3 & 4 would score 10 points). Higher counts can therefore be achieved by either a number of high scores (patent classes – specialisation) or a large number of lower scores across a wide spread of patent classes). • The Y-axis is a count of the number of patenting classes in which industries achieved a 3 or 4 on the measure of technological strength (i.e. niche or core strengths on Patel and Pavitt’s measure). Therefore, the industry in the example above would score 2 points. Therefore, for an industry to score in the top right hand quadrant it must have a high total count of patent activity (x-axis) and it must have a high number of patenting niche or core strengths. Computing and instruments are industries with high R&D intensive and similar capital intensities but with a low technology depth and breadth3 (on the basis of patents). They both have a relatively low aggregate patent specialisation count (X-axis) with a moderately low degree of concentration – only 3 patent classes received a 3 or a 4 (Y-axis). The result for computing might be because the original data was restricted to the major companies and/or because patenting in the computer industry only delivers limited benefits – due to the speed of technological change, and is hence not as commonly adopted as other industries such as pharmaceuticals. Pharmaceuticals, by contrast is an example of an industry highly focussed on technological depth. It has five patent classes with a score of three or four but it has a lower aggregate specialisation than instruments or computing. Eighteen of the pharmaceuticals industry’s 21 patenting points came in niche and core strengths, so the assessment can be made that technological depth is important to its competitiveness. Companies in the aerospace industry have the same number of technological specialisations as the

3 Ozman (2007) can differentiate between information technology, telecommunications and semiconductors, all of which have quite different scores.

0

1

2

3

4

5

6

0

5

10

15

20

Electrical

30

Build

Computers

Rubber Textiles

Food

35

Paper Metals

Mining & petroleum

Motor veicles

Patenting strength

25

Drink

Instruments

Pharma

Chem

40

45

Aircraft

Machines

50

6

Fig. 6.6 Patenting specialization: concentration vs. spread Note: original data from Patel and Pavitt 1997

Patenting concentration

7

122 Clustering Internationalisation

6.3

Imports, Science and Scale

123

pharmaceuticals industry, but have an accumulated patent specialisation score of 46. This is more than twice that of pharmaceuticals. The chemicals industry has the largest spread of niche and core strengths (7 patent classes), whilst patenting by firms in the motor vehicles industry have neither a high degree of patenting breadth or depth. This analysis helps us to understand that some industries such as aircraft and aerospace,4 as an extreme example, rely on a large number of technological fields. This presumably reflects their production system that relies upon a very large number of components that need to be integrated. Such information sheds new light on the nature of aerospace and pharmaceuticals, which were both classified as sciencebased by Pavitt (1984), but which have high and low import dependencies. This analysis is not wholly satisfactory, as it is based on very old data. It is, however, one way with existing data (Patel and Pavitt 1997) to separate industries on the basis of technological specialisation based in depth or breadth.

6.3 6.3.1

Imports, Science and Scale Import Intensiveness

In terms of the scale of cross border interdependencies, the first section of this chapter identified clusters in the EU and the OECD models which relied most heavily upon imported materials, components and services. This is possible because input-output modelling of inter-country production flows makes it possible to calculate the percentage of output that is spent on imports. A list of high import-intensive national-meso clusters from the two models reveals that each model produced somewhat similar results, indicating that the original input-output data is relatively consistent. A wide range of cluster types appeared in these lists. Given the literature which, links new knowledge and clustering (see Chap. 3), it is perhaps surprising that R&D intensity reveals little of the key characteristics of high import clusters, as such, clusters are spread reasonably evenly across R&D categories (Table 6.4). In contrast, using a modified Pavitt taxonomy reveals that the scale intensive category can account for nearly four times more clusters than other categories for the EU model and more than twice any other category for the OECD model (Table 6.5). This result, inturn, highlights the deficiency of such taxonomies which are based on only one dimension. For example, science based industries might require scale and some scale based industries are also R&D intensive. Three clusters, namely ‘office and computing machines’ (science based) and ‘radio, television and communications equipment’ (specialised supplier) and ‘motor vehicles’ (scale intensive) exhibit a combination of capital, R&D and import intensiveness.

4

See Brusoni 2001 for a discussion of this industry, its knowledge bases and firm boundaries.

124

6

Clustering Internationalisation

Identifying R&D and capital intensiveness also separated resource based scale industries and capitalised industries with a technology base. This evidence suggests one direction in which industry taxonomy analysis could be developed further is to allowing for distinctions between different types of labour intensive, resource intensive, specialised supplier, scale based and science based activities to be derived from available data. Important anomalies remain, however, notably; pharmaceuticals and instruments with a high capital ratio and low import values and aerospace which is high on the R&D and import scales but with a low capital ratio (for the available data period). Intuitively, the difference would be expected in the nature of their product complexity (components) and this was seen to some degree in the analysis of technological complexity. This analysis suggests that imports for particular clusters are a major source of technology, albeit embodied in components.

6.3.2

Implications for Theory, Research and Policy

The results presented here on the changing structure of intermediate imports with their particular relevance to high technology manufacturing needs to be explored further. Case study analyses based in qualitative information gathered from businesses and focused on tracking particular value systems across product lines, industries and jurisdictional borders are needed. In so doing it will be essential to tie in geography, technological specialisation and end product complexity. There is some evidence that as technology complexity rises, so to does both specialisation and the degree of dependence on innovation networks (see, e.g. Frenken 2000). Although, it is not wise to draw too heavily on these findings for the design of innovation policies there are some warnings. A considerable emphasis of writings on cluster policies has been on developing high technology and often ICT support strategies. For example, a recent publication of the Organisation for Economic Cooperation and Development states: ‘Several of the more recent cluster/regional specialisation programmes were born from science and technology policy. They promote collaborative R&D to support growth of the most promising technology sectors in regions where these sectors are concentrated. Albeit in theory spatially neutral, in practice such policies often focus on specific geographic areas where key institutions, researchers and firms are clustered’ (2007a, p. 12).

As Martin and Sunley (2003) suggested, cluster policies are no panacea and the findings presented in this chapter would reinforce this. The OECD (2007a) suggests that cluster policies should be regionally and sectorally specific. Given, the indicators of internationalisation presented here and OECD (2007b) some other tentative suggestions are possible. Lower technologies should not be ignored and governments need to be very clear about the specific knowledge and other competitive advantage of their regions. Further, depending upon the specifics, it may hamper regions and firms if imports are discouraged for industries with particularly complex technological systems.

Chapter 7

Cluster Complexes: Auto Production

alongside hollowing-out (in the sense of relocation from Europe) there is an ongoing and parallel process of inward investment into Europe. It is in the precise balance of these two forces that the structure of the automotive components industry in Europe is currently taking shape (Sadler 1999, p. 118).

7.1

The Significance of Automotive Production Systems

The auto industry is often seemingly perceived by many governments as something of a growth engine and significant politically. Presumably, this is due to its linkages to many other industries including petroleum, metals, and electronics, because of its large scale employment and because of its connection to national consumer culture (see, e.g. Paterson 2000). As such, there is a huge literature on the international industrial structure and the technological systems of transport related industries for many countries. Global auto production has been analysed from many different angles and often by authors with an intimate knowledge of the activities in individual countries or companies. A key theme of this literature has been the continuing international re-organisation (see Dicken 2003a, b) of the industry and technology. As Chap. 6 revealed, scale-based and some leading technology sectors were reliant on an internationalised production system, one of which was the auto industry. Thus, it is logical, using the capabilities of the input-output models developed here to explore the spatial architecture of the industry from the 1970s onwards. Although, the analysis developed here cannot have the depth of existing material it nevertheless contributes a perspective that isn’t available elsewhere. The focus therefore is not to reiterate the wealth of information already available but to relate the information on global structures to other information on production geographies, clustering and innovation systems. To establish the context for the discussion that follows, the first section of the chapter deals with the export structure of the motor vehicle industry.

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_7, © Springer-Verlag Berlin Heidelberg 2009

125

126

7.2

7

Cluster Complexes: Auto Production

Background: A Traditional Perspective of Exports

The spread of car production to a wide range of countries makes it worthwhile establishing which regions have a significant role in the production system. We will start the analysis with 1990 data, just before many of the larger recent changes started. World Trade Organization data (Fig. 7.1) reveals that exports by countries within the EU to other countries within the EU (intra-exports) are nearly three times larger than external exports and approximately four times greater than the USA’s exports to all other countries in 1990. Moving forward to 2005, intra-European exports still represented a major share of global cross-national trade (Table 7.1). Trade between the USA and Canada was the next largest grouping of trade between countries. One feature of the data in Table 7.1 to note is that, according to the WTO, the fastest growing trade interactions between 2000 and 2005 are intra-Asian. In the previous five years (1995– 2000), the fastest growing flows were from Latin America to North America (WTO 2003). It should be noted that the WTO division of countries has Mexico, which is a part of the North American Free Trade Agreement (NAFTA) arrangements, within the group of Latin American countries. It would seem likely, then, that the Free Trade Agreement has (unsurprisingly) facilitated the rapid development of auto production in Mexico (see Fig. 7.2).

Fig. 7.1 Automotive exports 1990 (share of world exports) Notes: (b) Includes significant exports from processing zones. Source: based on data from WTO (2003) Table IV.54. http://www.wto.org/ english/res_e/statis_e/its2003_e/its03_bysector_e.htm WTO (2003)

7.2

Background: A Traditional Perspective of Exports

127

Table 7.1 Major regional trade flows of automotive products (US$b)

Intra-Europe (EU 25) Intra-North America Asia to North America Europe to North America Intra-Asia Asia to Europe Europe to Asia

Value 2005 $B.

Annual percentage change

2005 391.3 156.8 74.3 51.3 47.2 37.8 21.7

2000–05 11 2 6 12 17 12 12

Source: WTO (World Trade Organization) 2006 Table: IV: 62

Mexico (b) Korea, Republic of European Union (15) Extra exports Czech Republic (b) United States Canada Hungary (b) Poland Thailand Slovak Republic

Regional blocks and countries

Turkey China (b) Argentina South Africa Russian Federation (c) Brazil Slovenia Australia Norway Singapore domestic exports Switzerland Romania Europen Union (15) Intra exports Japan –8.00

–6.00

–4.00

–2.00

0.00

2.00

4.00

6.00

Fig. 7.2 Changes in world export share of automotive products 1990–2001 Notes: (b) Includes significant exports from processing zones. (c) Includes Secretariat estimates Source: WTO (2003) Table IV.54 http://www.wto.org/english/res_e/statis_e/its2003_e/its03_bysector_e.htm

Middle East

Africa

South & Central Am

Asia

North America

Europe

Destination regions

0.0

5.0

10.0

15.0

20.0

25.0

30.0

% of world trade

7

Fig. 7.3 Share of world exports in auto industry products by supplier and destination regions Source: Based on data from WTO (2006) Table IV.64

Europe North America Asia Japan South and Central America

Supplier regions

35.0

40.0

45.0

128 Cluster Complexes: Auto Production

Other regions C’wealth of Independent States

7.3

The Evolution of the Inter-cluster Networking

129

Intra-Asian trade in auto products is still relatively small. The lack of cross border regional trade could, in part, be due, as Dicken suggests, to Japanese manufacturers re-creating separate supply chains ‘out of the necessity created by high levels of import protection in virtually all the East Asian countries, particularly those in South East Asia (notably Malaysia)’ (2003a, p.45). The difference between the absence of integration apparent in the auto industry in East Asia and the situation in the ICT and electronics industries, where Asia is the largest multi-country regional intra-trade block (see Chap. 9), is striking. Between 1990 and 2001, Mexico achieved very rapid growth in its share of world exports (Fig. 7.2), gaining about 4%. Korea gained 2% extra of world exports and the EU 15 around 1%. For the same period, Japan and intra-European Union exports dropped as a share of world automotive trade by more than 6% each. Intra-regional trade is a dominant feature of trade and it represents the major share of a country’s cross border activities (see Fig. 7.3). Europe, as with the other major trading regions are principally supplied through intra-regional flows. Apart from Japanese trade with North America, intra-regional trade dominates trading destination preferences.

7.3

The Evolution of the Inter-cluster Networking

Inter-regional input-output modelling based on net output flows facilitates an analysis of the connectivity of industries in a number of different ways. The two used here are first data on the actual contribution of imports to total industry output. The second is the ‘mapping’ of linkages which represent more than 10% of the imported share of output. As becomes apparent, this perspective changes the overall look of trade flows (compared to exports).

7.3.1

Auto Activities 1970

The input-output data for 1970 is limited but enlightening Fig (Table 7.2). The first aspect of the data to notice is that imports represent a relatively small share of production. Apart from Canada and the Netherlands, imports are generally less than one quarter of all inputs into the production process. The second feature to notice is that the category of the Rest of World, which within this modelling work represents all regions not specifically identified in the model, represents a high share of the imports. The Fig. 7.4 shifts the focus to the international architecture of imports. Within the data for 1970, there is a clear structure. Apart from the imports from the Rest of the World (ROW) there are two clear sub-systems, one based in Europe, centred on Germany and another sub-system, a bi-nodal structure for the Asia–Pacific– Americas (APAs). In this book such large scale configurations of basically distinct

130

7

Cluster Complexes: Auto Production

Table 7.2 Imports as share of output in 1970 Country

Imports share of $1 of output

Imports from the Rest Of the World

Share of imports outside the nine %

Australia Canada Denmark France Germany Japan Netherlands UK USA

0.264 0.431 N/A 0.090 0.157 0.091 0.439 0.155 0.077

0.061 0.060 N/A 0.055 0.099 0.065 0.190 0.089 0.034

23.1 13.9 N/A 61.1 63.1 71.4 43.3 57.4 44.2

network systems of major hubs and second tier nodes will be called exo-nets1 (external-networks) and the system of linkages for any particular country its cluster complex. In the European exo-net only the UK has linkages to APAs. There are no linkages going the other way. Within Europe, Germany is a clear hub as a supplier to the other economies in the system. Interestingly, it has a bilateral relationship with France. In the other exo-net the USA buys from both Canada and Japan, Japan buys off the USA, Canada buys off both the USA and Japan. It should be noted that the figure presents the source of components relative to each production structure, so although the USA represents a very major market for Canada from the American perspective it is a significant but not a dominant supplier. Australia buys from both Japan and the USA. The latter case already shows a preference for Japan, even though the Australian industry at the time was essentially European and American owned.2

7.3.2

Auto Activities 1990

Twenty years later, in 1990 the international system as seen through the lens of the same eight economies had changed remarkably little. For Canada, Germany, the Netherlands and the USA, the share of imports in the production process grew by between four and eight cents per dollar. The Australian production system drew on only half a cent more imports per dollar of output and the Japanese drew on half a cent less imports. Both the UK and France experienced dramatic rises in the use of imports (Table 7.3).

1

This language is an attempt to get away from descriptions such as triads etc. The more standard term fits export profiles as we have seen in the earlier analysis but doesn’t work as well when the analysis focuses on interdependencies. 2 Maxcy (1965) indicates that car assembly companies present were Ford, British Motor Corp, Chrysler, Volkswagen and General Motors-Holden (with nearly 50% of the market). Fujimoto (1999) notes that Toyota entered the Australian production system in 1968.

Fig. 7.4 1970: inter-cluster auto production networks (linkages above 10% of imports). Notes: The arrows point in the direction of the financial flows in accordance with input-output traditions. Products flow in the reverse direction. Data for Denmark was missing as it was combined with another industry category

7.3 The Evolution of the Inter-cluster Networking 131

132

7

Cluster Complexes: Auto Production

Table 7.3 1990 Aggregate imports comparison Country

Total imports share Imports - Up or of $1 of output Down over 1970

Imports from the Share of imports Rest of the world outside the nine

Australia Canada Denmark France Germany Japan Netherlands UK USA

0.269 0.502 N/A 0.242 0.200 0.086 0.503 0.316 0.155

0.080 0.077 N/A 0.143 0.127 0.053 0.227 0.148 0.065

0.005 0.071 N/A 0.152 0.043 −0.005 0.064 0.161 0.078

29.61 15.39 N/A 59.25 63.46 61.95 45.14 46.88 44.2

Interestingly, for many countries their dependence on countries outside of this group of countries (i.e. the ROW) did not markedly shift. France decreased dependence on the ROW by about 1.9% while the USA did not move at all, and Germany (0.4%), Canada (1.5%) and the Netherlands (1.8%) marginally increased theirs. The outliers were Australia which de-specialised, increasing dependence on the ROW by 6.5% and Japan (9.5%) and the UK (10.5%) which concentrated their import sources. However, these changes did not alter the architecture of the system (Fig. 7.5 on page 131).

7.3.3

EU 15 1995

The evidence presented above suggests that the European Union countries are highly dependent on Germany. This can be explored further with the use of the inter-regional modelling of interactions within the EU 15. The data for the 15 countries of Europe combines all transport related industries into a single classification, aggregating aerospace, motor vehicles, shipbuilding and other transport (railways, etc). These different industries are separated in the OECD data. The degree to which any transport national meso-cluster in Europe relies upon imports is hugely variable. Within the EU model, transport had the most internationalised production system, ranked by the number of times it appeared in a list of the top 5 internationalised clusters for each nation. Table 7.4 provides the share of output (as a percent of each Euro) imported by each member of the EU 15 in 1995. The larger economies import notably less per Euro than most economies, but curiously, France imports less than Germany. Greece imports less than any, probably due to an under-developed industry structure. Countries that appear in the tail of the list require more than 40% of their output to be imported to support production. The geographic structure of imports is presented in Fig. 7.6. Germany is clearly a ‘hub’ economy with all but two economies (Ireland and Luxembourg) being dependent on it for at least 10% of their component imports. France is a second tier supplier, particularly with links to nearby economies. All

Fig. 7.5 1990: inter-cluster auto production networks (linkages above 10% of imports) Note: Arrows point in the direction of the financial flows, products flow in the reverse. Data for Denmark was missing

7.3 The Evolution of the Inter-cluster Networking 133

134

7

Cluster Complexes: Auto Production

Table 7.4 EU 15 transport equip mfg: import dependency Imports – Cents per $ of output Greece France Germany UK Ireland Italy Spain Finland Sweden Portugal Denmark Netherlands Austria Belgium Luxembourg

0.18 0.20 0.24 0.30 0.30 0.32 0.35 0.36 0.39 0.39 0.43 0.45 0.50 0.63 0.76

Fig. 7.6 1995 EU 15 transport mfg sector import dependencies

third tier suppliers (UK, Spain, Italy and the Netherlands) have relationships with adjacent economies. The above diagram shows clearly why the structure of relations matter, as they differ between countries and further highlights the inadequacy of concepts such as supra-national innovation systems or Rugman and D’Cruz’s (1993) double diamond. The analysis of the spatial structure of interdependencies reveals a layered structure to the inter-industry relations for the transport clusters in European countries.

7.3

The Evolution of the Inter-cluster Networking

135

This structure of production and trading hierarchy has not gone unobserved, although the input-output modelling presented here brings new clarity to the discussion. Hudson observes: The automobile production system is seen, by some, as organised within European-wide networks, encompassing a three-fold hierarchy of regions, qualitatively differentiated in terms of their role in the production system. R&D and high level and knowledge-intensive competencies are increasingly concentrated in the core, centred on Germany, as routine production, especially of lower value models is increasingly dispersed to the eastern and southern peripheries. This emergent hierarchy ‘is based upon the cumulative competencies of the actors, the density of networks of relationships and proximity to the seats of power where strategic decisions are taken … distributing other activities over space’ (Bordenave and Lung, 1996, p. 320). However, creating such a regionally hierarchical Europeanised production system is complicated in at least two ways. First, ‘national champions’ still dominate in some national markets (Bordenave and Lung, 1993; Hudson and Schamp, 1995). Secondly, supply chains are being extended beyond Europe (Sadler, 1999) (2002, p. 14).

Lung (2003) attributes this layered, poly-central pattern to the history of national development prior to the processes of economic integration, with Paris, Turin and Munich part of the centrifugal dynamic and the centres in Belgium and Northern France being more peripheral. The input-output data would suggest that Italy is also more peripheral than economically central. The trend in R&D expenditures in Europe’s auto industry, as analysed by Leoncini and Montresor, supports this tiered structure and suggests a co-evolutionary process where different layers have differential rates of investment in technology. The progressive increase of the (total manufacturing weighted) R&D expenditure, and of the revealed (by patents) technological advantage of the German motor vehicles, as opposed to, respectively, a less substantial increase and a decrease in France and Great Britain (1999, p. 34).

A question that arises from the literature and this analysis but which largely goes unanswered regards the linkage between the technological specialisation of clusters and the products that are being supplied between clusters (i.e. the technological links between clusters).

7.3.4

Auto Activities 2000

Alongside Germany, which has just been shown to be a major producer of transport components, Diehl (2001) observes that the other major producers of auto components are the USA and Japan. Together these three countries account for two thirds of all auto parts. The input-output modelling for 2000, which incorporates data for 22 partner countries plus Mexico as a supplier but not an importer due to the unavailability of data, supports this analysis. The addition of more countries over the 1970 and 1990 inter-regional models, adds valuable information, but again highlights the overall stability of trade structures (Fig. 7.7).

7

Fig. 7.7 Auto industry integration 2000 Even in 2000, two exo-nets are clearly apparent. The United Kingdom’s purchases off the USA and Greece’s dependence on Japan, are the only links between exo-nets (at the chosen level of significance). The other observation to make is that the two exo-nets have an important structural difference. Countries in Europe tend to have more geographically specialised cluster complexes, when compared to those of the APAs exo-net. Germany, is the primary hub, with few secondary linkages

136 Cluster Complexes: Auto Production

7.3

The Evolution of the Inter-cluster Networking

137

In contrast, the APAs exo-net has two strong interdependent hubs. The Asia– Pacific countries, excluding China source from both Japan and the USA. Brazil and Canada specialise their sourcing on the USA, Canada having weakened its link to Japan during the 1990s, probably as a result of Japanese producers expanding production within North America. As well as its links to Japan, America has significant links with Canada and Mexico. The production structure of these networks is quite interesting (Fig. 7.7). Imports as a share of each dollar of production in Australia and Japan and the United Kingdom shifted little from 1990. Canada, Germany, France and the USA imported slightly more (between 5 and 8 cents per dollar) over their figure for 1990. However, the Netherlands became considerably more dependent on imports. Table 7.5 highlights the general shift in the requirement for imports in the production process. Between 1970 and 1990, the process was relatively slow but it sped up during the 1990s. As of 2000 the average requirement for the group of countries that represents most of the original core group – the G73 (plus Australia and the Netherlands but excluding Italy) is close to the average for a wider group of mainly OECD economies (plus China and Brazil). Table 7.5 2000 aggregate imports and change over previous years

Australia Austria Brazil Canada China Czech Republic Denmark Finland France Germany Greece Hungary Italy Japan Korea Mexico Netherlands Norway Poland Spain Taiwan UK USA

3

Imports as share of $1 of output

Change over 1970 y2000 minus y1970

Change over 1990 y2000 minus y1990

0.34 0.61 0.20 0.57 0.20 0.67 0.42 0.40 0.29 0.31 0.44 0.74 0.30 0.10 0.31 N/A 0.81 0.34 0.40 0.56 0.32 0.32 0.19

0.07

0.07

0.14

0.07

0.20 0.15

0.04 0.11

0.01

0.01

0.37

0.31

0.17 0.11

0.01 0.04

The G7 Canada, France, Germany, Italy, Japan, UK and USA existed between 1976 and 1998 see http://www.g7.utoronto.ca/ accessed 9 November 2007.

138

7

Cluster Complexes: Auto Production

A rising share of these imports has been from production centres beyond the core group of countries. As Table 7.7 reveals most of the increase in ROW imports occurred during the 1990s (except for the UK). Despite these changes, the availability of the new data (OECD 2006)4 facilitates an analysis that captures the majority of activity, as seen in Table 7.8. While interdependencies between the core-group of eight countries in the modelling represented a high share of imports between 1970 and 1990, this pattern broke down during the 1990s. Calculated on a basis consistent with the 1970 and 1990 classification of the ROW, this category accounted for 18 cents of imports per $ of output in 2000. However, despite the diversifying supply base for the core group of eight countries just noted, the increase in model size from eight countries to 22 reduced the share of external trade for the core countries to less than the endogenous capture available in 1970. The average for all 22 countries in the model was a little more at 11.6 Thus, as the models endogenously capture a very significant share of the international economic activity in motor vehicle production, the diagrams presented above reveal the majority of important trade links. Table 7.6 Total imports requirement (% of $1 of output)

Average

1970 (core eight countries)

1990 (core eight countries)

2000 (core eight countries)

2000 (22 countries)

0.21

0.28

0.37

0.40

Table 7.7 Imports from the ROW – country consistent basis

Australia Canada Denmark France Germany Japan Netherlands UK USA

Imports from ROW Total – 2000 (consistent with earlier periods)

Change 1970 – 2000 (cents in $)

Change 1990 -2000 (cents in $)

0.13 0.11 0.23 0.17 0.20 0.06 0.27 0.17 0.10

0.07 0.05

0.05 0.04

0.11 0.10 0.00 0.08 0.08 0.07

0.03 0.07 0.01 0.04 0.02 0.04

Table 7.8 The capture of the ROW in the model – average of ROW imports

Average

1970 (ROW – 1990 ((ROW – 2000 –(ROW – external for external for core external for core eight) eight) core eight))

2000 – Core eight ROW beyond model

2000 - (ROW for all 22 countries)

0.081

0.070

0.116

0.115

0.181

4 It was noted in Chapter five that the OECD has made available domestic and imports I-O tables for more countries than could be developed into an inter-regional model (due to trade data access constraints) at the time of writing.

7.4 Techno-Organisational Context of Auto Production

7.4

139

Techno-Organisational5 Context of Auto Production

The analysis so far is informative vis-à-vis the overarching structure of industry evolution and patterns of change, but is less useful in helping us understand the dynamics behind those changes. Therefore, this section focuses on the corporate, technological and geographic issues behind the architecture of production in motor vehicle production. The analyses in this chapter of the evolution of the international networks of motor vehicle national-meso clusters reveal a number of features about the scale and spatial configuration of cross-national interdependencies. In general, the scale of bilateral linkages is small, in line with previous inter-regional modelling experience, and thus the majority of links were not shown on the earlier diagrams.6 For every country, however, there are concentrated inter-industry linkages that extend beyond national borders. These pathways of traded interdependencies tend to fit with previous trade analysis that suggests geographic proximity is important for international trade. The spatial structure of each set of cluster linkages investigated in this chapter is different. For many countries value is still more local than imported, but this has been weakening significantly as imports rise with the share of imports averaging 40% of output in 2000.

7.4.1

Clusters, Complexes and System Hierarchies

This book rests on the huge extant body of literature that analyses the characteristics of regional clusters, but the analyses here can only extend to the interactions between national clusters. However, it is possible to show that these national-meso clusters (based in systems of interdependency) are strongly related to the clustering of economic activity in particular places. Table 7.9 reveals that the auto industry is highly concentrated into just a few locations in each national economy. For each of these countries, a limited number of regions represent a very high share of total employment in the motor vehicle industry. One curiosity of this data is the consistency in the scale of the largest cluster. National employment shares of the largest regional (states and provinces) clusters account for 25.3% (France), 28.6% (Germany), 28.2% (Japan), 28.2% (UK) and 23.6% (USA). It would be reasonable to speculate that between first tier and second tier national clusters there would be considerable trade. It would also be interesting to see research on the scaling of clusters.

5

The term is used here to refer to the technological architecture of the final product, the process technologies and important organisational characteristics that give a context for development patterns. 6 As was noted in Chap. 5 changes to the benchmark statistic will considerably alter the look of the figures.

140

7

Cluster Complexes: Auto Production

Table 7.9 Employment in motor vehicle production centres (circa 2000) Regional clusters (names include regional codes for Europe)

National employment

Provincial employment

Provincial share of national employment

52,000 Australiaa, b Victoria 27,000 51.9 South Australia 13,500 26.0 30,678 Austria (2001)c at3 Westösterreich 13,917 45.4 Belgium 1999 54,446 be2 Vlaams Gewest 43,706 80.2 171,002 Canadab Ontario 113,000 66.1 7,604 Denmarkd, b 253,885 France b fr4 Est 64,232 25.3 fr2 Bassin Parisien 52,196 20.6 fr1 Île de France 47,031 18.5 767,096 Germanyb de1 Baden-Württemberg 219,626 28.6 de2 Bayern 165,994 21.6 dea Nordrhein-Westfalen 89,958 11.7 178,573 Italyb itc1 Piemonte 76,505 42.8 itc4 Lombardia 31,283 17.5 916,432 Japane Aichi 258,269 28.2 Shizuoka 109,718 12.0 Kanagawa 70,493 7.7 28,009 Netherlandsb nl4 Zuid-Nederland 15,567 55.6 9,1041 Polandd, b pl22 Slaskie 20,011 22.0 164,559 Spainb es5 Este 63,062 38.3 es2 Noreste 39,561 24.0 79,002 Swedenb se0a Västsverige 41,180 52.1 243,025 UKb ukg West Midlands 68,416 28.2 ukj South East 24,952 10.3 1,115,200 USAf Michigan 262,959 23.6 Ohio 136,400 12.2 BEA South East USA 252,935 22.7 a Australia – Productivity Commission (2002) b Data from Eurostat 2006 c Canada – National stats from Industry Canada 2004 with data for Ontario for the year 1999 from Fitzgibbon et al. (2003). d Denmark and Poland – some of the regional data was withheld by Eurostat for confidentiality reasons e Statistics Bureau of Japan (2007) data for 2006 f USA data from the Bureau of Economic Analysis (2005) downloaded 11 August 2005. BEA South East Region includes Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia

7.4 Techno-Organisational Context of Auto Production

141

Beyond the intra-national clusters, nations form only a few major supply chains. In turn, this structure of linkages, preferences some clusters over others so that there are ‘hub’, as well as second and third tier clusters.

7.4.2

Geography and Organisation

The analysis of international networks and national clustering, above, reveals that international production has a particular architecture and national clusters are dominated by specific local clusters. However, so far, the missing piece in the analysis has been the geography of corporations. The foregoing analysis emphases production and trade geography over the political economy of tariffs and other barriers or the inter-related geography of where the auto manufacturers establish their assembly plants and how suppliers cluster around. The political economy of protection (see, e.g. Maxcy 1965, Sturgeon and Florida 1999 and Dicken 2003a) and regional subsidies (see, Klier 1999) will, for the most part, be left to one side in this analysis. This is simply because there is not the space here to consider it. On the other hand, it is quite valuable to remember that ‘corporate geography’ is too easy to leave out of the analysis (see Dicken and Thrift 1992) on system geography. The analysis so far has shown two exo-nets and the structure of individual national cluster complexes at the international scale. Interestingly, analysis of the sub-national clustering of economic activity also reveals some patterns of significance. It is commonly known that North American auto manufacturers are concentrated in the USA’s Mid-West (Michigan and States to the south) as well as the Canadian Province of Ontario to the north. Japanese FDI into North America in this sector accelerated in the late 1980s and concentrated in this same geography (see Mair et al. 1988). During the 1990s, this pattern of co-location in the States of Michigan, Indiana, Ohio, Kentucky and Tennessee by both Japanese and European organizations largely continued (Klier 1999). The latter analysis emphasises that although new assembly plants may locate outside of this ‘corridor’ (Mercedes in Alabama), the supplier plants are likely to remain inside the corridor to take advantage of important infrastructure. Thus the location of the network of suppliers is apparently driven by both a requirement to be near assembly plants due to just in time production and the structure of interstate highways, which ‘apparently allows for frequent deliveries to multiple customers from a single supplier plant location’ (Klier p. 31). Thus, Japanese and European firms both invest in U.S. production. This, however, doesn’t advance us towards an understanding of the structure of the exo-nets. One piece of evidence is that large firms like Toyota do seem to substitute investment for trade. However, suppliers do not follow the same trend (Head and Ries 2001). There is less evidence on European investment7 in the USA but there is some general

7

Evidence in Pries (2003) suggests that European investments by BMW and Damlier-Chrysler within the USA have not been in traditional USA auto states.

142

7

Cluster Complexes: Auto Production

evidence on the reverse. FDI investment into Europe by American corporations is large and has been focused on the UK and Germany in particular. Schoenberger (1990) suggests that this is driven by both concerns over protectionism and with the evolution of European market integration; a desire to be on the inside. The geographic foci of American firms might also suggest differing strategies like those suggested by Head and Ries, suggesting that investment in the UK might have been by suppliers (see Figs. 7.4, 7.5 and 7.7) leading to backward linkages to the USA, while investment in Germany has been by lead firms.

7.4.3

Reconfiguration

The foregoing analysis reveals that the core structural architecture of international production of motor vehicles appears to shift very gradually, with a few exceptions such as the UK. Lung (2003), who has examined the location decisions for car assembly plants in Europe between 1991 and 2005, has revealed that many of the new plants have been, and will continue to be concentrated in Germany. His tally of car plants opening and closing in Europe was: • • • • • • •

Austria: gained 1 plant; France: 2 shut down and 2 opened; Germany: lost none and gained 6 new car plants; Italy: 4 plants shut down and only 1 opened; Spain: 2 plants closed but 1 new plant opened; Sweden has had 3 plants shut and only gained one; and The UK has had 4 plants shut, with 2 opening.

This pattern of concentration in Europe ties well to the analysis presented so far in that it also suggests that Germany has been strengthening its hub position in Europe since the early 1990s. Alongside the production geography, the innovation geography has been evolving slowly along similar lines. Technological and industrial specialisation patterns exhibiting a ‘high degree of stability in the face of change’ in the motor vehicle industry is achieved, according to Leoncini and Montresor (1999, p. 34) by a high degree of internal technological change. This is supported by Cantwell and Iammarino’s recent analysis of patenting profiles across time of various regions in Europe, which included motor vehicle related technologies. Baden Wurttemberg had the strongest technological position relative to the world, followed at a considerable distance by the Ile de France and then Lombardia and the South East of the UK (2003, p. 133). Their study also reveals that although the Ile de France, the UK’s South East and Lombardia all slipped backwards in their technological specialisation, Baden Wurttemberg remained stable (2003, p. 144). Therefore, the evidence of the historical trend appears to be consistent at the organizational, technological and the macro input-output level of analyses.

7.4 Techno-Organisational Context of Auto Production

143

In contrast, it is apparent that new geographies of production are emerging in Central and Eastern Europe with a tone in some of the literature that this might be a general trend. Dicken, in particular, draws attention to this theme. As the same supplier observed: Quite clearly, our goal is ultimately to develop manufacturing plants in lower wage cost countries. We will develop a supply base around those. In simple terms, the route we are taking at the moment is that we take a UK assembly and a UK supply base and the first thing we move is the assembly, then we need to develop the infrastructure around the new locations. (Supplier company interview) (2003a, p. 25).

How this spatial structure will change with the introduction of the 10 accession countries into the EU in 2004 and two more in 2007 is still debatable. Lung (2003), Hudson (2002) and Dicken (2003a) have all discussed this with the common interest in how much of Europe’s production will move east. In the view of Lung (2003), the core areas of auto specialisation are not threatened but production on the periphery is prone to shifting more easily. Such a view is also supported by the long time series evidence presented in this chapter. On the other side of the Atlantic, analysis is not as straightforward as for Europe because statistics (estimates) for trade between States of America are not readily available. It would be valuable to be able to go beyond just the interactions of national clusters alone and model a multi-state system for the USA to monitor the internal production geography of the USA as the core apparently shifts away from Detroit. The Economist has argued that Detroit as a production centre is in trouble as the Japanese manufacturers are establishing competing production activities in other locations across America, particularly in the southern states. ‘In the past seven years Detroit’s share of the American market has slid from 73 to 63%. If SUVs [sports utility vehicles], pick-ups and the like are excluded, the big three’s share of the passenger-car market is already under half’ (2003a, p. 11). What happens to the Detroit cluster may be important to the Canadian cluster as it is co-located near Detroit and the two are highly interdependent on each other for parts and components. Whether supply lines will (are) stretch(ing) to the new locations in the Southern USA from Canada or whether these new production locations will drive further growth in Mexico are important questions for the countries of NAFTA. On the basis of current data, (see Fitzgibbons et al. 2003) the declining share of Detroit is not leading to a decline in the absolute level of output from the Windsor-Toronto auto corridor. To the contrary, this Canadian cluster actually grew strongly during the 1990s and 2000s.8 At this point, the example of the global auto industry well highlights the difference between the innovation systems literature’s emphasis on local technological competencies and the approach of authors concerned with the shift of production away

8 See Rutherford and Holmes (2008) for an interesting discussion of clusters versus global corporate organisations. However 2008 was a year of turmoil for the auto industry with sharp oil price rises and a resulting sharp fall of in truck and SUV sales (Knowledge@Wharton 2008) and then the international financial turmoil (see Gurría 2008).

144

7

Cluster Complexes: Auto Production

from traditional locations, particularly to low cost developing economies. Malerba, for example, on the one hand, observes that ‘automobiles, with few innovators, geographically concentrated with local knowledge boundaries are associated to technological regimes characterized by high cumulativeness at the firm level and a system type of knowledge with some tacit components’ (2002, p. 260) focuses on the geographic concentration of knowledge development. The growing trend of major global manufacturers to develop their car platform strategies on the basis of increased product modularisation,9 is important for the issues of innovation and coordination. However, modularisation is seemingly not breaking down the overall geo-centric nature of assembly – as might be assumed. Sadler argues that ‘component manufacturers were less dependent on home (national) markets than assemblers, and had developed a wider spatial spread of operations even before 1990, a difference which became more marked in the period up to 1995’ (1999, p. 118). On the other hand, Sturgeon and Florida suggest that suppliers follow assemblers to new markets to be involved in the design process, even whilst manufacturers in high income countries seek to import components from low cost countries (see, 1999, p. 83). Alliances and the popularity of mega-mergers within the motor vehicle industry have facilitated access to technology without necessarily increasing cross-border traded interdependencies through product licensing (etc) (see Dicken 2003, Sturgeon and Florida 1999 and Howell and Hsu 2002). This confusing picture from the detailed analyses of economic and political forces both encouraging concentration and simultaneous feeding fragmentation send us back again to the macro data on patterns of structural evolution. There is a general trend of multi-national regional production systems to be integrated across the nearest available high-income border (Korea10-Japan, Japan-Australia, France-Germany, Canada-USA, Mexico-USA). However, the spatial structure of inter-country interdependencies is not limited to such bilateral flows or even the multi-country triad. Second at the aggregate level fragmentation (occurring at the periphery) and agglomeration (occurring at the core) is proceeding slowly. Lastly, although local technological capability does seem to be important, limiting innovation systems analyses to national or sub-national systems excludes the emerging shape of the world economy.

9

Sako (2003) as well as Takeishi and Fujimoto (2003) both have good discussions of ‘product hierarchy’and ‘product architecture’ (Sako 2003, p. 250). It is apparent that from the mid-1990s product modularisation increased, especially through the construction of new plants by the large German manufacturers Volkswagen (in Brazil, Czech Republic and, Eastern Germany) and Damlier Chyrsler in France and the USA (Takeishi and Fujimoto 2003). While the ‘suppliers develop and assemble subassemblies’ (Takeishi and Fujimoto 2003, p. 261), these suppliers ‘surround’ new assembly plants (p. 262). 10 For analysis of the auto industry development in the Republic of Korea see Park (2003).

7.5

7.5

Implications and Conclusions

145

Implications and Conclusions

Whatever else the picture presented in this chapter indicates, it is not one where the ‘world is flat’. Production is fragmenting and presumably specialising, but during the 30-year snapshot available, here the development pathway is very structured geographically. Two weakly related systems (in terms of significant traded interdependencies) of cluster complexes emerge, which is in contrast to the usual language of the economic triad. There is apparently a clear dichotomy between a European exo-net focused on Germany and an Asia–Pacific–Americas (APAs) exo-net (Canada, USA, Mexico, Japan, Korea, Taiwan and China11 and Australia). The European exo-net, here shown to be strongly dependent on Germany also has a series of weak ties to economies such as France, Italy and the UK. The Asia–Pacific–Americas system is different from the European system in that rather than being strongly tied to a single economy it is constructed by a series of cluster complexes that taken together form the exo-net. Canada is heavily reliant on components from the USA, the link between the USA and Japan is a binary network of mutual dependency and Australia is dependent on production in Japan and the USA. It was noted above that the pattern of a limited number of strong links generally goes against what is often found in multi-regional input-output studies. While the APAs exo-net is distinct, the Euro complex has linkages across the divide to the USA (exporting to the UK) and Japan (selling to Greece). The absence of a significant level of links between Germany and either Japan or the USA is the most striking feature of this analysis. Germany dominates the European industry, but it has not used its national production base to develop big export markets in the other geographic regions. This is, of course, a question of corporate strategy. Japanese firms have established production in North America and it is known that German firms have done likewise, perhaps ameliorating their need to import from home. However, the evidence for the dynamics of European ‘transplants’ is weaker than for their Japanese counterparts. The twin emphases of the existing literature, the largely stable technological and skill patterns and the changing supplier networks, remains a critical issue in the literature. It does seem, however, that examining cluster architectures can be useful. We have observed that ‘hub’ economies’ have kept their position, even strengthening over time, while some peripheral clusters have weakened. One exception to this general rule, which is indicated by some secondary evidence, is the case of Detroit. As valuable as this information is, cluster connections are not the whole story. Motor vehicle producers have adopted a number of paradigms including flexible production, lean production, just in time production, and more recently the greater modularisation of components. The latter has involved not just greater fragmentation of production sources but also the increased involvement of suppliers in design.

11

For commentary on China’s auto industry see e.g Sit and Liu (2000).

146

7

Cluster Complexes: Auto Production

Cluster analysis and theory cannot be left purely in the realms of geography but must be more linked to the techno-organisational setting of particular production systems. Further, it is clear from the trend of some production moving to Central and Eastern Europe that macro-dynamics (politics, economics, etc) can be highly influential in the ‘clustering’ of industry. Finally, more research needs to be done on the role of physical and political geographies and even jurisdictional borders play in the structuring of innovation systems. Rather than suggesting policy proposals, the implications of these findings lie elsewhere. Without better awareness of the pattern of fragmentation, system integration and the role of high value services (design), geography and industrial change, it seems likely that public money will be wasted on protection or attraction policies that do not have the hoped for positive impacts.

Chapter 8

Cluster Complexes: Civil Aerospace

8.1

Aerospace: The Techno-Organisational Context

In contrast to auto assembly which has become less concentrated geographically, but retains a significant number of original equipment manufacturers in a mass production system across thousands of supplier firms,1 aerospace assembly is different. It requires the integration of a massive number of components2 (and absolute quality assurance), with low unit output, and has become heavily concentrated (corporately). Chapter 6 revealed aircraft and aerospace companies spend a high proportion of their turnover on R&D and invest it across a very diverse range of technological fields. Niosi and Zhegu characterise global aircraft and aerospace production as having: large dominant firms, high barriers to entry, increasing returns and high upfront R&D costs, world markets for products, and inertia: large plants imply high sunk costs. Attractors are large systems integrators, not universities, government laboratories or other institutions (2002, p. 3).

To manage the required scale, risks and specialisation, the system integrators are now concentrated with just four players in civil aerospace, but with a developed internationalised production system. While design is a core competency of system integrators such as the assemblers and first tier suppliers, increasingly some development costs are pushed out to suppliers (see Nolan et al. 2007). Increasingly the Primes are turning to loosely coupled networks of suppliers and the modularisation of production to enable them to take advantage of fast moving areas of science and technology while maintaining effort in slower moving areas (see Brusoni et al., 2001). To manage this, large system integrators require a set of different competencies, allowing them to design, build, outsource and assemble. As Prencipe noted of the aero engine maker, Rolls Royce:

1

Nolan et al. (2007) suggests that in the early 1990s each lead assembler firm had a supplier network of 30,000 firms but by 2000 this had dropped to 5,000. In contrast Boeing in 2005 had 1,200 supplier firms. 2 Aero-engines embody up to 40,000 components (Prencipe 1997: 1266).

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_8, © Springer-Verlag Berlin Heidelberg 2009

147

148

8

Cluster Complexes

it could be argued that vertical integration, total design capability, system integration capability, and intelligent customership have allowed Rolls-Royce to develop and maintain an all-round ability concerning the gas turbine engine. (1997, p. 1275).

Such sets of competencies are important in a sector with the strong evolutionary trajectories visible in aircraft designs (see Frenken and Leydesdorff 2000), but where technological discontinuities disrupt industry development from time to time. However, management of the in-house competencies and the loosely coupled networks is not without problems. Both Airbus and Boeing have experienced difficulties with their recent new product development. Airbus delayed production of its new A380 by two years due to problems with wiring3 and Boeing’s 787 has also been delayed. In the latter example the challenges of project management involving the integration of competing software systems are highlighted by Holmes: ‘suppliers such as Smiths Aerospace, Honeywell International and Rockwell Collins are fierce competitors that have been corralled into being teammates on the 787 program. As each vendor struggled to get its software to talk to the others, competitive juices boiled … “I’m used to knowing who my enemy is every step of the way” says one supplier familiar with the matter. “But now I’m working with the enemy’…” (2006, p. 40).

It is common to see each case of developing new aircraft designs such as the A380 and the 787 as a ‘betting the company’ strategy. Thus, the barriers to ‘entry’ have also proven to be barriers to the development of new aircraft. Post World War II there were 34 companies in the business of producing commercial aircraft (Golich 1992, p. 904) in Europe and the USA. By 1980, this had declined to five companies through failures, mergers, acquisitions and for a number of North American companies, in particular, the choice to specialise in military aircraft. Twenty years later, after the merger of McDonnell Douglas and Boeing in 1997 (see Olienyk and Carbaugh 1999), there were just two major assemblers (Boeing and Airbus) and two regional jet producers (Bombardier and Embraer). Until 2001, BAE Systems produced the Avro-RJ4 a four engine special performance jet for up to 112 passengers. Underlying this concentration has been the continuing escalation of costs of producing new aircraft. Golich (1992, p. 905) provides a guide to development costs, although not providing actual dollar estimates. On a scale that has five major points, The DC8 was measured at about 0.5, the DC10 was closer to 1.5 and the DC model in the design phase was estimated at more than 5. One of the more recent new planes, the Airbus A380 is estimated to have cost €12b or USD $17b in development costs.5 Pritchard and MacPherson (2004) comment that such estimates could be at the low end, just as the official figures for the Being 777 were probably understated. They also note that Boeing’s new 787 (although smaller) will probably be just as expensive to develop as the new Airbus aircraft. This can be placed in the context that it may take 600 unit sales to break even on the development costs (Olienyk and Carbaugh 1999, p. 64). 3

See e.g. http://www.spiegel.de/international/0,1518,438408,00.html accessed 20 November 2007. http://www.baesystems.com/ProductsServices/ra_avro_rj.html accessed 13 November 2007. 5 Cutler and Neely (2007). 4

8.2

Cluster Networking

149

The combination of industry cost drivers, production strategy and the politics of global aerospace that require local purchases generates a system of international purchases and corporate alliances (see Nolan et al. 2007, Macpherson and Pritchard 2003 and Pritchard and MacPherson 2004). Therefore, the analysis of the changing architecture of cluster networks which follows reflects changing production, technical and political realities.

8.2

Cluster Networking

In the automobile production system, it was possible to observe that the technoorganisational systems influenced the geographic pattern of international clustercomplexes. In the aerospace industry, this context is again informative for helping us to understand the evolving structure of international production.

8.2.1

Aerospace 1970

In the auto production system (Chap. 7) it was possible to clearly observe two international exo-nets that had very few linkages between them. The exo-net structure remained largely stable across a period of 30 years, although new countries were added to the modelling. A similar international structure is not visible in the aerospace production system. In the 1970 data, we observe that all countries (for which data is available) had strong linkages with the USA (see Fig. 8.1). The backwards linkages to the USA account for a high percentage of their imports. Although the Rest of the World (ROW) accounts for a high percentage of imports for France, Germany, the United Kingdom and the USA, these countries, in general, require very a small share of inputs from overseas sourcing (see Table 8.1). One feature that might surprise some readers is that, although it is a very small absolute amount, the USA was importing more that 10% of its international inputs from Japan in 1970.

Table 8.1 Imports per unit of output 1970 Imports Imports from ROW cents in $1 output Australia 0.255 0.03 Canada 0.227 0.03 Denmark N/A N/A France 0.146 0.05 Germany 0.178 0.07 Japan 0.245 0.04 Netherlands 0.492 0.10 UK 0.123 0.06 USA 0.045 0.02

Share of ROW % 11.76 13.22 34.25 39.33 16.33 20.33 48.78 44.44

8

Fig. 8.1 Aerospace cluster networking 1970 NB. Arrows point in the direction of financial flows – goods and services flow in opposite direction

150 Cluster Complexes

8.2

Cluster Networking

8.2.2

151

Aerospace 1990

By 1990, imports as a share of production had risen (Table 8.2) for most countries in the model (excluding Australia). Interestingly, although the absolute level of content being sourced from countries exogenous to the model had also increased, there was either a decline or a negligible increase in the share that the ROW represented (again with the exception of Australia). The dynamics of the Australian industry may be due to the Government off-sets programme that was in operation to require some work on aero components be conducted in Australia with the purchase of major commercial aircraft (See DITR 2003). In 1990, of the eight possible national meso clusters studied here, aerospace had a higher imported value added than the relevant auto industry in five countries. Those cases where the import percentage was less than those for auto counterparts were the USA (the largest national system), Canada (with a niche industry) and Australia (with a negligible aerospace industry in 1990). The spatial dimension of these connections indicates the strength of the USA in 1990 (see Fig. 8.2 over page). It seems little changed between 1970 and 1990.

8.2.3

Aerospace 1995–2000

Due to the data limitations of the OECD latter datasets, the 1995 and 2000 analysis need to be considered together. Unfortunately, due to the statistics reporting policies of the European Union, input-output data is not published for aerospace. In developing the 2002 release of the OECD database (2002e), the OECD estimated some of the values for aerospace for a number of countries. At the time of preparing this analysis this has not been prepared with the most recent release (OECD 2006). The picture that emerges for 1995 (Fig. 8.3) and 2000 (Fig. 8.4) is complex. Although the USA remains a primary global hub, Germany and France have emerged by 1995 as a second tier of production. Further, for 1995 this second tier is primarily created by trade links within Europe, the trade flows between the USA and China from France are notable. Table 8.2 Imports per unit of output 1990

Australia Canada Denmark France Germany Japan Netherlands UK_ USA

Imports per $ output 0.16 0.25 N/A 0.33 0.21 0.30 0.59 0.32 0.09

Imports share 1970–1990% −0.10 0.02 N/A 0.19 0.03 0.06 0.10 0.20 0.04

Imports from ROW cents in $1 of output 0.05 0.03 N/A 0.12 0.08 0.04 0.12 0.10 0.04

ROW Imports share 31.2 13.7 N/A 36.5 36.2 13.5 19.9 30.3 46.7

ROW share 1970–1990 19.4 0.4 N/A 2.2 −3.1 −2.9 −0.4 −18.5 2.2

8

Fig. 8.2 Aerospace cluster networking 1990

152 Cluster Complexes

Cluster Networking

Fig. 8.3 Aerospace cluster networking 1995

8.2 153

8

Fig. 8.4 Aerospace cluster networking 2000

154 Cluster Complexes

8.2

Cluster Networking

155

By 2000, some more changes are evident, even with the limited data available. Not unexpectedly, the USA remains central. France and Germany, probably retain their second tier position. However, two more countries join this second tier. The UK has built stronger trade links with Canada and Australia and the Republic of Korea. Japan continues to maintain its position as a supplier to the USA and in line with other production networks it is a source of imports for other East Asian economies.

8.2.4

Aerospace Import Changes

Table 8.3 compares import shares. European countries appear to have stabilized their international sourcing during the 1990s. Japanese production became less dependent on imports while Australia, Canada and the USA became more so. Given the scale of the USA’s system, its change is of particular interest. Although ROW imports, external to the original core countries modelled in 1970 and 1990 expanded during the 1990s (Table 8.4), the model was able to capture most of the change (Table 8.5). Table 8.3 Comparison of import shares 1990–2000 1990 1995 2000 Change Australia 0.16 0.41 0.37 0.22 Canada 0.25 0.30 0.36 0.11 Denmark N/A France 0.33 0.35 0.02 Germany 0.21 Japan 0.30 0.24 0.23 −0.08 Netherlands 0.59 0.58 −0.01 UK 0.32 0.34 −0.66 USA 0.09 0.11 0.17 0.08 Table 8.4 Core countries – ROW on a time consistent basis ROW 1995–2000 ROW 1990 Change 1990–1995–2000 Australia 1995 0.08 0.05 0.034 Australia 2000 0.05 0.05 0.002 Canada 1995 0.05 0.03 0.014 Canada 2000 0.05 0.03 0.019 Denmark 1995 Denmark 2000 0.18 France 1995 0.12 0.12 −0.003 France 2000 Germany Japan 1995 0.04 0.04 0.000 Japan 2000 0.05 0.04 0.008 Netherlands 1995 0.24 0.12 0.122 Netherlands 2000 UK 1995 0.20 0.10 0.099 UK 2000 USA 1995 0.04 0.04 0.004 USA 2000 0.06 0.04 0.024

156

8

Cluster Complexes

Table 8.5 Core Countries – ROW – external to model 1990 1995 2000 Change Australia 0.05 0.07 0.03 −0.02 Canada 0.03 0.03 0.03 0.00 Denmark N/A N/A 0.12 N/A France 0.12 0.08 N/A −0.04 Germany 0.08 N/A N/A N/A Japan 0.04 0.03 0.04 −0.01 Netherlands 0.12 0.19 N/A 0.07 UK 0.10 0.17 N/A 0.07 USA 0.04 0.03 0.04 0.00

Thus, although the database upon which the modelling is constructed is incomplete, the model does account for a high percentage of economic activity in the aerospace production system for the countries that are included.

8.3

The Beginnings of the Future of Aerospace Geography

Sometime during the 1990s, the global configuration of civil aerospace production began to change. Obviously, the success of Airbus6 during the 1990s has been influential in maintaining and enhancing European capabilities in commercial aerospace production. Unlike auto production where European and Asia-Pacific – Americas exo-nets were apparent, in aerospace there is only one exo-net even from 1970, although the individual cluster complexes vary. In 1990, America was unmistakably a first tier production system. Only businesses in Germany supplying France, the UK and the Netherlands form a second tier. From 1970, Japan and Canada both managed to sell into the U.S market. Evidence in Niosi and Zhegu (2002) suggests that apart of Canada’s trade strength could be in business jet parts where Canada has a technological and industrial specialisation. In Figs. 8.1 and 8.2, it is possible to see the very strong backward links between countries, mostly with America, but also with the ROW. Given that it is the Dutch,

6

Airbus started as an ad hoc consortium during September 1967 to develop commercial aircraft with a memorandum of understanding between Great Britain, France and Germany. The MOU designated a lead contractor and their lead structural component – France was to lead on airframe design and Britain was to develop engines. Each country agreed that their national airlines would be encouraged to purchase from Airbus. Airbus Industrie was formed in December 1970 (see Hayward 1987). During the 1990s, Airbus Industrie was owned by Aerospatiale S.A. (France), Daimler-Benz Aerospace, British Aerospace and Construcciones Aeronauticas S.A. (Spain) (see Olienyk and Carbaugh 1999). In 2000 the corporate character and governance of Airbus was again transformed with the agreement of European Aeronautic Defence and Space Company (EADS) and British Aerospace (BAE) in 2001.http://www.baesystems.com/Newsroom/NewsReleases/2001/press_120720011.html accessed 15 November 2007. In 2006 BAE sold its 20 per cent share. BAE website http://www.guardian.co.uk/ business/2006/oct/04/baesystemsbusiness accessed 15 November 2007.

8.3 The Beginnings of the Future of Aerospace Geography

157

French, Germans and British that have strong links to the ROW, it is probable that a number of the suppliers to these countries are in Europe. It is possible to speculate that the dual link between France and Germany apparent in the 1990 data (Fig. 8.2) is the nascent Airbus Industries global production network. The transition between 1989, when Airbus (with aircraft assembly in France) had one third of the global market,7 and 2000 when Airbus reported it had gained more orders8 that its rival Boeing indicates important changes. Within the architecture of aerospace production, the emergence of the UK as a second tier supplier, visible in the cluster networking map for 2000 (Fig. 8.4) is most surprising. Two of the countries that have purchasing links with the UK also have supply contracts with Airbus for wing components (see Table 8.7 below) which is led by UK based firms. Therefore, perhaps in the lead up to the manufacturing work by Korea and Australia there was a transfer of design, equipment and components from the UK to Korea and Australia. Notwithstanding the performance of European economies, total global aerospace production is still dominated by the USA, which accounts for about one half of all turnover and the European Union for about one third (STAR 21 2002, p. 14). At the start of this Chapter, the techno-organisational context of aerospace production was introduced, particularly a few of the drivers of consolidation. These changes have driven productivity improvements which have enabled twice as many commercial jet aircraft to be produced in 2000 as in 1980, albeit with large-scale job losses. For example, America’s employment in this industry declined by a third between 1989 and 2000 (Botham et al. 2001) and the USA was not alone: the main centres of European production (the UK, France and Germany) failed to maintain employment. ‘In France employment has fallen by 10% since the early 1980s and in Germany it has been static. The situation in Great Britain and Scotland is shown in Fig. 2.1. Continuing a steady reduction which began during the 1970s, GB employment has declined by over 40% since 1981 (81,300 jobs). The equivalent figures for Scotland are almost 50% and 5,500 jobs. Much of this decline occurred in the early 1990s’ (Botham et al. 2001, p. 7).

Even after falls in employment, the UK’s aerospace cluster still employs twice as many as the third biggest (Germany), (see Table 8.6) What then is the likely scale and spatial structure of aerospace production and innovation systems into the future? European-based activities have downsized employment but increased their share of the value of global production. To keep pace with the USA, projects in the EU are increasingly designed on a pan-European basis (STAR 21, p. 15). Airbus has centred its assembly in Toulouse France, just as Boeing centres its assembly activities on Seattle, Washington. Recent research, however, appears to have shown little interest in the spatial structure of links in aerospace, concentrating more on the management of networks or the national innovation systems. Botham et al. do, however, make this interesting observation:

7 http://www.centennialofflight.gov/essay/Aerospace/Airbus/Aero52.htm accessed 15 November 2007. 8 http://seattlepi.nwsource.com/business/286106_boeing22.html accessed 15 November 2007.

158

8

Cluster Complexes

Table 8.6 Turnover and employment in aerospace production Turnovera in billion Euro 116.6 80.6

USAc EU

Employmentb (’000) 588.6 435.5

Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden UK Japan Canada Otherse Total

Employment in EU countries (‘000)

3.8 7.4 1.3 1.1 104.4 74.8 4.0 4.3 39.2 540 11.2 3.8 19.6 12.7 147.1 13.0 15.9 10.3 236.3

31.3d 83.6 102.0 1241.0

Total 435,539 Notes: a consolidated turnover b at year end c excluding sales and employment not directly pertaining to aerospace d includes company staff only directly related to aerospace production (i.e. figure not comparable with other regions) e estimate, PR China and CIS not included Source: AECMA (Fig 35 2002, p. 38) and AECMA 2002b (Fig. 24, p37)

‘The success of Airbus will benefit the UK industry. However, it should be noted that the major UK companies are (and long have been) important suppliers to Boeing. At least some increase in output and employment due to the growth of Airbus could be at the expense of suppliers to Boeing’ (2001, p. 10).

Although the degree to which UK based firms can become a part of the successful Airbus supply system would seem to be a critical issue for future growth, the authors quoted here are more concerned to investigate the input-output based multiplier effects of Scottish local aerospace activities for Scotland. They reveal a disregard for considering the demand side of a cluster’s production and how it fits within a broader product system. This is, however, not surprising as demand is not a popular topic in either economic geography or systems of innovation research, which is identified by Bresnahan et al. ‘‘demand is important for growth’. In a debate overemphasizing supply side factors like agglomeration economics and external effects, however, this remark is often overlooked’ (2001, p. 843).

Production of a fully-fitted section of the A380’s central fuselage in three pieces. Fuselage shells. Tail planes Composite assembly structures.

Airframe (fuselage) The forward, central and rear upper shells. Cargo doors The centre wing box - the ‘heart’ of the aircraft. Nose section and forward fuselage sections.

Design Alenia Aeronautica - fuselage centre section. High lift systems, as well as flaps, tracks/fairings and sheet metal parts. Engineering design centre Wing design. Various system design projects. Systems Avionics, Air-conditioning, etc.

Components. Advanced Metals and Comonents Fasteners. Aluminium. VSMPO-AVISMA provides aircraft makers with a wide-range of materials – titanium, aluminium, magnesium alloys and steel, including sponge, ingots, slabs, extruded Ti sections, panels, seamless and welded tubes and forgings. Alloy metal parts.

Table 8.7 Airbus A380: geography of major component production

Nordenham, Bremen, Stade and Hamburg.

Varel, Germany.

Japan.

Turin and Naples. Nordenham.

Hamburg, Germany and from Italy. Japan Nantes, France. Meaulte, France.

Multiple companies and sites across the USA and Europe.

(continued)

Assembled at Saint Nazaire France. Saint Nazaire, France. Hamburg.

Delivered to Saint Nazaire. Spain.

All sites

Turkey. Austria, UK. Russia.

Rome. Bremem, Germany. Wichita, Kansas, USA. Filton, UK. Multiple companies across the USA.

Delivered to

Production Location

8.3 The Beginnings of the Future of Aerospace Geography 159

Engines Rolls Royce. Engine Alliance 50/50 joint venture between GE Aircraft Engines and Pratt & Whitney. Assembly Belly fairing and Horizontal tail plane. The wings. Assembly of the complete rear fuselage. Front and rear fuselage sections. Final assembly line Cabin furnishing and painting. Source: information derived from Airbus (2007)

Undercarriage assembly Radial tires. The main landing gear.

Table 8.7 (continued) Components. Vertical tail plane, pressure bulkhead and flaps. Carbon fibre components for the horizontal tail plane, skins, spars and elevators. Wing sections Fixed leading edge lower wing panels. Wing bottom panel. Wingtip fences. Wing ribs. A380 wings.

Toulouse, France. Toulouse, France. Toulouse, France. Hamburg, Germany. Customers

Hamburg, Germany. Toulouse, France. Hamburg, Germany.

The final gear integration will take place at the Goodrich facility in Toulouse, France.

Puerto Real, Spain. Broughton, UK.

UK.

Japan. The parts are being produced by Goodrich’s landing gear division at its manufacturing facilities at Oakville, Ontario; Cleveland, Ohio; Tullahoma, Tennessee; and Krosno, Poland.

Delivered to Filton UK. Delivered to Filton UK. Delivered to Filton UK.

Malaysia Korea Sydney, Australia. Filton, UK. Broughton, UK.

Delivered to Toulouse, France.

Delivered to

Production Location Stade, Germany. Illescas Spain.

160 8 Cluster Complexes

8.4

Implications and Conclusions

161

In the case of Airbus, the supply chain is particularly spatially extended. Table 8.7 identifies the production sites for the new Airbus9 super-jumbo A380, an aircraft with a passenger capacity of 550 that entered into service late in 2007. The complexity of integrating this number of major components from this number of sites across Europe and the world (from as far away as Australia) has required Airbus to develop an integrated logistics plan and the construction of specialist transport equipment. An Airbus report states: ‘Transport of the aircraft sections to the final assembly line in Toulouse includes a mix of sea, river, road and air transport. An itinerary for the oversized loads has been developed to move sections from Airbus sites across the world. A huge roll-on, roll-off sea vessel will be used to take components on the first stage of the journey, by sea, from Airbus sites in the UK, Germany, France and Spain to the French city port of Bordeaux. Specially designed barges will then carry the components on the penultimate part of the voyage, along the Garonne River, from Bordeaux to the river harbour of Langon. Here the aircraft components will be transferred to road trailer to continue the final part of the journey to the final assembly line’ (2004, p. 3).

Thus, the emerging input-output derived evidence on networks appears to be supported by the more direct evidence on component purchasing behaviour and production geography. This is encouraging for continuing to develop this form of multi-spatial systems of innovation analysis. The scale of the change shown in Figs. 8.3 and 8.4 indicate that the transformation underway in global aerospace will result in a spatial architecture that remains hierarchical but is less and less uni-polar. However, it is unlikely to evolve to a bipolar system similar to motor vehicle production. This analysis highlights, that although pathways of traded interdependencies tend to confirm that geographic proximity matters, the particularities of production systems also matter. One other speculation may be warranted. Modularisation in the auto industry does not seem to have significantly altered the shape of cluster networking across the period 1970–2000. However, similar project and technology management trends within aerospace maybe having a larger and more rapid effect on the global geographic architecture of knowledge, production and flows within aerospace. Local clusters are developing macro-modules such as engines or airframe sub-assemblies for final assembly elsewhere and these patterns do seem to be appearing in the structure of trade networks.

8.4

Implications and Conclusions

In aerospace, the USA has been the dominant specialist and hub economy for the period 1970–2000. All cluster complexes in aerospace involve the USA. However, changes visible in the analysis presented here raises interesting questions regarding the deep processes that drive economic development in this sector.

9 Boeing, is less forthcoming over its supplier relations, whereas Airbus provides a very significant amount of detail.

162

8.4.1

8

Cluster Complexes

Innovation Theory and Analysis

The analysis of the aerospace industry presented here begins to raise an important question for innovation theory. For some time now, a key debate has been the degree to which R&D is being internationalised. This is typically understood to be a question over the degree to which multi-nationals are moving their research facilities offshore from their home country. However, this misses the bigger picture where complex technologies are being independently developed for components and then integrated into larger systems and product platforms. Take as an example the case of Rolls Royce. It invested its own capital into researching and developing the engines for the aerospace industry.10 The engines are the company’s final product but they are required to be designed and integrated into the larger final product of an aircraft. This is a case of national and internationalised R&D for a product platform that is itself internationally integrated. This is an important difference between the current perspectives on local clusters, which are perceived as local knowledge and value systems producing for international markets. The argument here is that while local knowledge contributions and accumulation is important, they must be understood as part of international value systems.

8.4.2

Policy Matters

As Boeing11 and Airbus continue to diversify their purchasing arrangements, push out to suppliers some of the costs of design, and take advantage of technological competencies new niches will emerge. McQuire comments: ‘Japanese subcontractors now make more of the aircraft by value than previously: for the 777 programme Japan’s workshare was 20% whereas for the 787 some 35% will be manufactured in Japan. For ongoing Boeing lines, Japanese manufacturers have assumed expanded responsibilities. Mitsubishi, for example, has long produced components for the 747–400 model; but in 2004 it took over complete fabrication of the wing centre section (Mitsubishi Heavy Industries 2004). Second, and far more significant, has been the move from production work to design. For most major American and European aircraft makers, ceding some production work was an expected, if sometimes cumbersome, way of clinching an order. What did not happen, until very recently, was to subcontract design of a major structural component of the aircraft’ (2007, p. 342).

The new niches will provide countries that have been on the periphery, but which nonetheless have advanced technological capabilities to capture some of the value of aerospace production. The case study of the Australian micro-cluster provided in

10 Total investment in research and development for Rolls Royce in 2006 was £747million (Rolls Royce 2007). 11 By way of example Boeing is adopting the Toyota Production System (see Cizmeci 2005) which is resulting in new opportunities for suppliers.

8.4

Implications and Conclusions

163

Chap. 4 exemplifies the opportunities. Thus while geographic outsourcing of manufacturing is occurring, for aerospace it is not moving to developing economies. Importantly, it is clear that shifts in value require continued investment over an extended period of time in skills, technologies and production by corporations and nation states. Finally, there needs to be policy realism. The patterns that have emerged here have evolved over long time periods as the technological demands of industry have increased. Further, the aerospace industry is influenced by various macro-dynamics driving airline demand, including: long-term economic growth (particularly Asia and Middle East), short to medium term volatility (e.g. September 11 2001) and fuel prices which themselves are influenced by global economic development.12

12

http://web.mit.edu/airlines/www/index2.htm accessed 16 November 2007.

Chapter 9

Cluster Complexes: Electronics and ICT

‘Trade data, over the years, show a greater integration of Asia. But it is a particular sort. It takes the form of an elaborate factory, or production network, where components move around the international system, but final product and the wellbeing of the network depend overwhelmingly on shipment to the U.S. In 1998 the US absorbed 35.4 percent of the region’s exports while providing only 17.5 percent of what the region imported’ (Cohen 2002, 22).

9.1

Introduction: Structures of Interdependencies

A key argument of the current book has been that at a macro level, clusters have a system of interdependencies just as businesses form local clusters through their business networks. From this position, the role, scale and spatial structure of interdependencies is of central interest. Intra-organisational interdependencies were classified by Thompson (1967), suggesting that they could be categorised as: • Sequential flows (where processing activities are conducted consecutively); • Pooled flows (where components need to come together in one place for processing); and • Reciprocal flows (where flows of activities move backwards and forwards between the different organisational units). The scale of a system clearly matters in how it might be classified. Motor vehicle production could be understood as either sequential (within the assembly plant) or pooled within a region. Obviously, Thompson’s concepts cannot therefore be entirely mapped onto value architectures,1 but essentially similar concepts can be utilised. In this context, it is a curious feature of the data that within the group of most internationalised cluster types, two fit the sequential product flow pattern at the value architecture level (petroleum and non-ferrous metals). Two could be 1 The term value architecture has been selected here deliberately to move away from the concept of the value chain which instantly brings to mind a linear production system that often ends with final manufacturing production and doesn’t consider their embodiment into services or as analysed here the lack of a linear system.

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_9, © Springer-Verlag Berlin Heidelberg 2009

165

166

9

Cluster Complexes: Electronics and ICT

considered pooled agglomeration patterns (motor vehicle assembly and aerospace.2 Note, however, that with the backwards I-O analysis there is also an apparent reverse effect with key hubs of agglomeration also then being significant suppliers to other locations. Finally, there are two that on the basis of case analysis fit the description of reciprocal flows (ICT and electronics). This chapter investigates the latter two, where the modularisation of production has generated significantly different international patterns from other industries. Some of the data contained in this chapter was previously used in Wixted and Cooper (2007), however the analysis here probes deeper into the emergence of clustering and inter-regional trade and formats the analysis slightly differently.

9.2

The Third Technological Revolution: ICT

Towards the end of the twentieth Century there was widespread commentary that information and communications technologies, including computing and electronics (often referred to as ICT) represented a new technological or industrial revolution, e.g. Begg et al., commented: … the problems that Europe faces in key areas such as growth, equality and employment are all related to its failure to take sufficient advantage of technological advances, particularly the ICT revolution…science-based industries, particularly those drawing heavily on ICT, have become the main driver of technological change and economic growth since the 1980s (1999, p 235).

The dot-com crash diminished some of the exuberance but not all the enthusiasm for the significance of ICT related economic activities.

9.2.1

Economic Growth and ICT

The growth and economic effects of information and communication technologies (ICT) has generated huge interest in academic, government and the private sector. Jorgensen et al. (2003) for example points to the productivity benefits derived from ICT activities. The OECD which conducted a large project on the economic growth effects of ICT concluded with largely positive sentiments, indicating that: Governments today are faced with a new economic environment. ICT has emerged as a key technology with the potential to transform economic and social activity and has led to more rapid growth in countries where the conditions for macroeconomic stability are in place.

2 However, the analysis in chapter eight of trade flows revealed that we may be witnessing the emergence of a global aerospace factory and reciprocal flows (rather than purely localisation) with the trend towards macro-modularisation of aerospace production, in the form of large section of airframes being assembled and then transported long distances for final assembly.

9.2

The Third Technological Revolution: ICT

167

While it is too early to say how important ICT’s transformations will be compared with those of previous innovations, like electricity, governments should nonetheless take action to manage adjustment and keep the social costs low. All governments can do more to exploit this new technology further, by accelerating its diffusion, providing the right skills and building confidence’ (2001b, p. 97).

On the other hand, Leamer and Storper (2001) are sceptical of the degree to which the new technologies will change the patterns of international trade and Smith points to the proponents of the benefits of ICT bundling it with unrelated technologies, commenting: ‘A related methodological issue concerns the link between technological complexity and interdependence, on the one hand, and economic effects on the other. We have noted several times that ICT is not one technology but many. At the same time, ICT is put to work in the context of major organizational changes, and often in the context of the application of other (unrelated) technologies. These facts are often neglected, with the effect that claims are made for the economic impacts of ICT which are not justified: the problem is to establish and use a framework which will allow interaction and multiple causality between technologies, organizational forms, and economic processes’ (2002, p. 42).

Alongside the proposition that information and communications technologies are useful to increasing productivity has been the argument that manufacturing such technologies can also be a spur to growth. The economic rise of places such as California, East Asia, Ireland, and Finland has been attributed to this strategy. The purpose here, then, is the same as in Chaps. 7 and 8 in that it seeks to look again at the evidence on clustering and scrutinise it from the perspective of the networks that have formed beyond the localised cluster. It is particularly important to note that prior to the OECD data for 1995 (2002e) and 2000 (2006), countries in East Asia,3 apart from Japan, were not included. This is significant to a discussion of the spatial architecture of linkages for these technologies. Their role in the global structure of ICT production is nevertheless discussed based on available information.

9.2.2

ICT Clustering

ICT and electronics related activities have been identified as a having a high propensity for innovative clustering due to the prevalence of knowledge spillovers (see Audretsch and Feldman 1996) and the need for deep labour pools. Thus, they are a

3

At some point in the future it is hoped that by using the Institute of Developing Economies’ Asian input-output tables for earlier periods, the scale and spatial configuration of linkages can be analysed in a way it deserves. Wixted (2005) has graphed the bilateral export pattern for all manufacturing industries in the OECD Bilateral Trade database for China, Hong Kong, Korea, Malaysia, Singapore, and Taiwan and Ireland as matrices; industry to destination. The charts cover the periods 1970, 1994 and 2000 to map the evolution of trade in both product and export partner specialisations across the period.

168

9

Cluster Complexes: Electronics and ICT

bundle of industries, activities and technologies that have been extensively studied for their cluster dynamics across a range of countries (see e.g. OECD 1999d and 2001c). However, there is little analysis that compares their economic geography across the globe. The pioneering cluster of ICT was in the Boston region around MIT, but for much of the history of the ‘industry’ it has been Silicon Valley (near Stanford University) that is understood to have been the dominant geographic location (see for e.g. Saxenian 1994 and Hulsink 2007). Audretsch and Feldman (1996: 632) identified 40% of computer industry innovations in the early 1980s as been generated in California. Recently, Feser et al., (2005) have identified 15 industrial complexes in the USA as combining value chains and high employment concentrations in ICT, the highest concentration occurring not in Silicon Valley but in Phoenix, Arizona. Undoubtedly, the growth of the computer industry globally, with the simultaneous growth of major computer brands many of which were at least initially based in California and the effects of this growth on the Californian economy (see Rhode 2001) have certainly built the perception that California is the example to emulate. ‘Follower’ clusters are often branded in a similar manner, ‘Silicon Fen’, ‘Silicon Glen’, ‘Silicon Wadi’ and the like. Despite continued technological leadership in the USA, production of ICT and electronics has largely continued the move west, first to Japan (beginning in the1960s) and then to other East Asian economies. As Table 9.1 reveals, Japan was already a major export centre in 1970 and rivalled the USA by 1980. However, note the rise of the other Asian economies following this period. Bernard and Ravenhill (1995) indicate that the exports of Asian economies are significantly orientated towards other Asian countries and to the USA. Ng and Yeats (1999 and 2003) show that the intra-Asian trade4 pattern is getting stronger not weaker.

Table 9.1 Computing and electronics exports (current US $ billion) East Asia ASEAN Japan Asia Total USA EEC7 Rest of World

1970

1980

1994

0.4 0.1 2.6 3.0 3.4 5.9 9.8

6.6 4.4 19.2 30.3 19.3 32.2 56.4

79.1 88.3 101.7 269.1 78.0 116.3 225.7

Source: Data from Sheenan and Tikhomirova 1996: 20

4

Of course it might also be the case that if there were reliable trade states for inter-State trade within the USA we might find a similar pattern.

9.2

The Third Technological Revolution: ICT

169

Even after the influence of their relatively close proximity is accounted for East Asian intra-trade must be generally classified as highly “intense”. In addition, the intensity of trade within the region increased markedly over both the full 1985–2001, and the shorter 1995–2001. For example, in 1985 only 40 percent of all East Asian bilateral trade flows were greater than expected, based on the countries’ shares in world trade, as opposed to 61 percent in 2001. Trade relations between most East Asian countries have been growing sharply in terms of their intensity and importance! (2003, p. 19).

With this background, it no surprise that the analysis presented in Chap. 6 revealed information, communication and electronic goods had a significantly internationalised production structure. However, the analyses of economic geography (production and employment) and trade profiles do not tell the whole story of international trade. The story of Apple’s IPhone5 is a useful illustration that while some value is captured by contract manufacturers, product designers and brand names can capture significant additional value. When leading firms are clustered into particular geographic places, such a finding suggests there is a need to research the degree to which particular places (clusters) are able to capture significant economic ‘rents’ from the creativity of the firms based there and the firms’ position and organisation of international production.

9.2.3

ICT Exports After 1990

Table 9.1 reveals that sometime between 1980 and 1994, economies in Asia other than Japan began to be significant exporters of computing and electronics products. In 1990, (Fig. 9.1) the major players in office and telecoms equipment were Japan, Europe, and the USA. Japan held more than 20% of world exports. Similar shares of exports were internal to the EU 15 and external to the USA, which had more than 15% of world exports. A group of nine East Asian6 economies each separately had around 5% or less, but together, not including Japan and not including re-exports by Singapore and Hong Kong, already held 21.7% of world exports. Canada’s export profile probably stands out as a traditional OECD country with a strong position in ICT products that may not have been expected.

5 Hesseldahl (2007) reports on a teardown analysis of the IPhone. ‘the cost of the materials used in the iPhone add up to about $200 for the 4-gigabyte version, which sells for $499 and about $220 for the 8-gigabyte version, which sells for $599. Their estimate doesn’t include costs of final assembly’ nor the amortised costs of design and R&D or profit. 6 Used in this chapter to refer to important East Asian economies: Singapore, Korea, Chinese Taipei, Malaysia, Hong Kong China, Thailand, China, Philipines and Indonesia.

170

9

Cluster Complexes: Electronics and ICT

Fig. 9.1 Office machines & telecom equip: share of world exports (1990)Source: WTO (2003) data from Table IV.46http://www.wto.org/english/res_e/statis_e/its2003_e/its03_bysector_e.htm

Between 1990 and 2002 (Fig. 9.2), the East Asian 9 economies (excluding Japan) exhibited a rapid rise in their share of world exports. Indeed the fastest growing country in ICT exports, China, captured an additional 7.25% of world exports between 1990 and 2002, apparently at the expense of Japan, in particular, which lost 13% from its share of world exports, falling to 9.7%. The USA, intra-EU trade, Hong Kong and Singapore also lost market share. By 2002 the East Asian 9 group held 36.2% of world exports (up by 14% of world exports), whilst internal EU trade was down to 18.5% and EU external trade was down to 9.4%. Taking a different perspective, Fig. 9.3 analyses country specialisations by comparing the ICT share of manufacturing value added. In this analysis, Finland and Ireland, as would be expected, are appreciably oriented towards these industries. In contrast, countries such as Australia have a very low degree of specialisation in this sector.

9.3

ICT Cluster Networks

171

Fig. 9.2 Office & telecom equip: change in share of world exports 1990–2002Source: WTO (2007) data from II.37Notes (a) or nearest year, (b) includes significant exports from processing zones, (c) figures refer to fiscal year, (d) includes Secretariat estimates

9.3 9.3.1

ICT Cluster Networks ICT 1970

Although in 1970, imports represented a small share of inputs into production of the larger economies, imports from outside the core set of countries already represented a surprisingly large share for most economies (the exception being Canada) (Table 9.2).

172

9

Cluster Complexes: Electronics and ICT

Finland (2001) Ireland (1999) (1) Korea (1999) (1) Japan (3,4) United States United Kingdom (2001) Canada

Countries

Hungary Mexico Austria Sweden Netherlands Denmark France Germany (1,3) Norway Portugal (1999) (1) Belgium (1) Czech Republic (1,3) Italy Australia (2000-01) Slovak Republic (1999) (1,3) Spain Greece (2001) (1,2,3) New Zealand (2)

0.0

2.5

5.0

7.5

10.0

12.5

15.0

17.5

20.0

22.5

25.0

Share of mfg value added

Fig. 9.3 Share of ICT manufacturing in total mfg value added (%) 2000 Source: OECD (2003b), see notes7

The diagram over the page (Fig. 9.4) reveals the network structure of Office equipment and Radio, Tv, and Communications equip industries for 1970.

7 Notes. 1. Data for rental of ICT goods (7123) are not available, 2. Postal services included with telecommunications services, 3. Data for ICT wholesale (5150) are not available, 4. Includes only part of computer related activities (72), 5. “Other ICT manufacturing” includes communication equipment, insulated wire and cable and precision instruments. “Other ICT service” includes wholesale and rental of ICT goods, 6. 1996 instead of 1995 for New Zealand, Norway and Portugal.

9.3

ICT Cluster Networks

173

Table 9.2 ICT & Electronics imports as a share of $1 of output Office and Computing Australia Canada Denmark France Germany Japan Netherlands UK_ USA

Radio, TV & Comms Equip

Imports

ROW

ROW share

Imports

ROW

ROW share

N/A 0.37 N/A 0.42 0.15 0.10 0.39 0.11 0.05

N/A 0.07 N/A 0.15 0.08 0.06 0.16 0.06 0.03

N/A 19.53 N/A 36.27 55.82 64.36 42.13 56.28 52.74

0.20 0.20 0.34 0.09 N/A 0.09 N/A 0.16 0.05

0.08 0.04 0.15 0.05 N/A 0.06 N/A 0.08 0.03

0.43 0.21 0.45 0.50 N/A 0.65 N/A 0.52 0.53

Figure 9.4 is constructed to facilitate the consideration of both industry classes of interest in this chapter simultaneously. The figure differentiates between where the connections are specific, or whether the statistically significant link is spread across multiple industries in the national-meso clusters (‘all industry’ groups). What is apparent from this analysis is that the three hubs seen elsewhere, of USA, Japan and Germany again re-emerge. However, the international structure is not as clear as with the previous production systems of auto and aerospace. A notable feature of this chart is that it reveals that specific supplier industries in both Japan and the USA individually met the test for significant trade links (10% of imports). Canada, in particular, is revealed as having a number of specialised trade linkages.

9.3.2

ICT 1990

Between 1970 and 1990, except for Canada, the UK and the USA, import requirements for office and computing equipment changed little, actually dropping in France. Radio, TV and Comms equipment import requirements grew by around 10% or less for most countries. The exception was Canada. Although the ROW represents a high share of import source, the percentage did not change greatly between 1970 and 1990, compare Tables 9.3 and 9.4.

9.3.3

ICT 1990–2000

The period 1990–2000 could be analysed either by production system type, that is ‘office and computing’ or ‘Radio, TV and Communications equip’ or by the trade structure for each hub cluster. As it turns out the architecture of the networks differs slightly between the two systems so the former approach will be adopted.

9

Fig. 9.4 1970 Cluster networks office & computing and radio, TV and communications Note: Arrows point in the directions of value flows – goods flow in the reverse direction

174 Cluster Complexes: Electronics and ICT

9.3

ICT Cluster Networks

175

Table 9.3 Comparison of imports as a share of requirements 1970-1990 Office and Computing Australia Canada Denmark France Germany Japan Netherlands UK USA

Radio, TV & Comms Equip

Imports 1970 Imports 1990

Change

Imports 1970

Imports 1990

Change

N/A 0.37 N/A 0.42 0.15 0.10 0.39 0.11 0.05

N/A 0.14 N/A −0.11 0.03 0.00 0.05 0.26 0.12

0.20 0.20 0.34 0.09 N/A 0.09 N/A 0.16 0.05

0.32 0.39 0.37 0.18 N/A 0.09 N/A 0.28 0.13

0.12 0.18 0.03 0.09 N/A 0.00 N/A 0.12 0.07

N/A 0.51 N/A 0.31 0.18 0.10 0.44 0.37 0.17

Table 9.4 Analysis of ROW imports - 1990 Office and Computing Australia Canada Denmark France Germany Japan Netherlands UK_ USA

9.3.3.1

Radio, TV & Comms Equip

Imports

ROW

ROW share

Imports

ROW

ROW share

N/A 0.51 N/A 0.31 0.18 0.10 0.44 0.37 0.17

N/A 0.12 N/A 0.12 0.10 0.05 0.17 0.16 0.09

N/A 23.19 N/A 39.41 56.10 51.75 39.92 43.15 53.30

0.32 0.39 0.37 0.18 N/A 0.09 N/A 0.28 0.13

0.13 0.09 0.16 0.09 N/A 0.05 N/A 0.13 0.07

39.93 23.13 44.77 50.58 N/A 58.16 N/A 45.03 55.16

Office and Computing Equipment

The first visual impression from Figs. 9.5 and 9.6 is that they are crowded with many more connections that were observed in the previous production systems. For Fig. 9.6 this is partly due to weaker connections (measured at between eight and 10% of imports) being introduced to the diagrams. This was done because of the number and variety of such linkages. In general, the structure of cluster complexes for individual countries includes more connections for office and computing equipment than for other systems. Further, these cluster complexes are more diverse than have been observed so far. For example, in the 1995 structure there is a two way linkage between the USA and Germany. However, the USA’s dependence on Germany had diminished by 2000. One interesting feature of the global architecture of office and computing supply structures is the emergence of second tier clusters. This feature was also observed in the aerospace production system. In the data for 2000, it is apparent that both the UK and France have become mini hubs. Mexico has two substantial buyers in the form of the USA and Canada, while Canada is not a major supplier to the USA, unlike other production systems that have been examined here. It is also worth pointing out that while China was a major buyer from Japan in 1995, there was a bilateral relationship in 2000 but both were 2nd order links.

9

Fig. 9.5 1995 Office and computing equip cluster networks

176 Cluster Complexes: Electronics and ICT

ICT Cluster Networks

Fig. 9.6 2000 Office and computing equip cluster networks

9.3 177

178

9

Cluster Complexes: Electronics and ICT

Table 9.5 Office and computing equipment imports - 2000 Australia Austria Brazil Canada China Czech Republic Denmark Finland France Germany Greece Hungary Italy Japan Korea Mexico Netherlands Norway Poland Spain Taiwan UK USA

Imports

ROW

ROW Share

0.4 0.5 0.2 0.7 0.4 0.8 0.4 0.8 0.4 0.5 0.2 0.9 0.6 0.2 0.6 N/A 0.8 0.5 0.3 0.5 0.6 0.5 0.2

0.2 0.1 0.0 0.1 0.1 0.2 0.1 0.2 0.1 0.2 0.1 0.3 0.2 0.1 0.2 N/A 0.2 0.2 0.1 0.2 0.1 0.2 0.1

44.1 27.6 24.9 9.7 32.7 27.2 35.2 30.1 33.3 33.2 30.3 33.4 29.2 45.4 27.4 N/A 27.4 32.3 38.1 36.0 20.0 38.2 35.4

It is a measure of the fragmentation of the production of office and computing equipment that, even with the increasing number of countries in the model, in 2000 the ROW represented a high share of imports (Table 9.5).

9.3.3.2

Radio, TV and Communications Equipment

The pattern of activity in the production of Radio, TV and Communications equipment is generally similar to office and computing but the density of connections is lower. This results in a structure of linkages for 1995 that somewhat resembles the two auto production exo-nets. Linkages in Europe predominantly point to Germany and as well as a system a linkages for the Asia-Pacific – Americas economies. What differentiates the two is that Japan and the USA are more connected to Europe, specifically France, the UK and the Scandinavian countries of Norway and Finland than in auto production. However, as has been pointed to before, the appearance of these diagrams (Figs. 9.7 and 9.8) is dependent on the level of statistical significance chosen, and that is why it is particularly valuable for Fig. 9.8 to include interdependencies just below the cut off. When this is done the architecture of the system becomes more complex. More European countries are revealed as having links to both the USA and Japan. There are also a couple of new links to the United Kingdom, Japan and China. The latter two again have a bilateral 2nd order supply system.

Fig. 9.7 1995 Radio, TV and Communications Equip cluster networks

9.3 ICT Cluster Networks 179

9

Fig. 9.8 2000 Radio, TV and Communications Equip cluster networks

180 Cluster Complexes: Electronics and ICT

9.4

Contemporary and Emerging Nodes of Global ICT

181

Table 9.6 Radio, TV and communications equipment imports - 2000 Error! No table of contents entries found.

Imports

ROW

ROW Share

Australia Austria Brazil Canada China Czech Republic Denmark Finland France Germany Greece Hungary Italy Japan Korea Mexico Netherlands Norway Poland Spain Taiwan UK USA

N/A 0.42 0.20 0.50 0.36 0.70 0.49 0.37 0.27 0.34 0.35 0.83 0.28 0.13 0.50 N/A 0.74 0.39 0.49 0.53 0.47 0.35 0.17

N/A 0.1 0.1 0.1 0.1 0.2 0.2 0.1 0.1 0.1 0.1 0.2 0.1 0.1 0.1 N/A 0.2 0.1 0.2 0.2 0.1 0.1 0.1

N/A 27.7 33.4 10.7 39.2 32.4 33.1 37.4 34.4 36.2 34.6 29.8 34.1 45.8 23.1 N/A 28.3 33.9 32.6 34.0 20.7 35.7 34.1

As a generalisation, import requirements for Radio TV and Communications equipment are less than for Office and Computing. Equally, the ROW represents a smaller share of imports in Radio, TV and Communications equipment than for Office and Computing (Table 9.6).

9.4

Contemporary and Emerging Nodes of Global ICT

The history of ICT and electronics industry development reveals a pattern of interaction between technology trajectories, the organisational structuring of relationships, industry policy and the slow accumulation of technological capability. This section explores first, the developments in the modularisation of electronic components and the organisation of design, manufacturing, assembly and distribution of merchandise in global production networks (GPNs). Second, this section briefly examines some of the evidence on the emergence of production centres in Asia and Europe (Ireland).

182

9.4.1

9

Cluster Complexes: Electronics and ICT

The Techno-Organisational Context in ICT

Any attempt to understand the emerging structure of production in the industries being examined here would be significantly lacking without at least a short review of the trajectories of technology and international business. In aerospace (see Chap. 8), there has been corporate consolidation and in the auto industry there has a trend to modularisation but little breakdown in the power of the OEMs. In ICT the modularisation of components, the de-verticalisation of the computer industry and the fragmentation of geographies of production do appear to be related (see Baldwin 2007). ‘Before the mid-1960s, computer designs were not modular. The first ‘truly modular’ computer design was IBM’s System/360, a broad compatible family of computers introduced in 1964’ (Baldwin and Clark 2000, p. 6). However, production at this time was integrated within IBM. In 1981, this changed with the decision to outsource components, with resulting changes to the industry structure. Field states: ‘As a result, the industry evolved along a path marked by the proliferation of disintegrated supplier firms subcontracting to PC makers, who, in turn, became dependent on the external capabilitiesof these suppliers and embedded in interfirm networks to build the product’ (2006, p. 127).

At the level of individual firms, the industry structure has evolved into one where a few businesses (flagship multinationals in the language of Ernst 2002) lead global production networks.8 Such organisations combine significant purchasing power, with requisite control over supplier activities. For example, Dell has exerted significant influence on inventory levels held by suppliers (Fields 2006). Some combine this power over the supply structure with system design and technology leadership (Ernst 2002). As these are ‘networks’ of businesses, opportunities arise for capacity and capability upgrading of national firms participating in global production, as indicated by Ernst and Kim: ‘Flagships place business orders and transfer valuable knowledge to local suppliers with only one objective in mind: to strengthen the competitiveness of their GPN. In response to intensifying global competition, the flagships’ outsourcing requirements have become more demanding. Typically, suppliers are selected by three criteria: a solid financial standing; high ratings on a quarterly scoreboard measuring performance in delivery, quality etc. and speed of response’ (2002, p. 1427).

Such production networks tie firms together but, in doing so, as firms are typically based in the systemic capabilities of regions, also link technological districts and in an agglomerated form are the basis of the cluster complexes observed earlier in this chapter.

8

Called variously, ‘global’ (Ernst 2002), ‘modular’ (Sturgeon 2003) and ‘international’ production networks.

9.4

Contemporary and Emerging Nodes of Global ICT

9.4.2

183

Contemporary East Asian Nodes

It has already been shown that economies in Asia have managed to achieve a remarkable growth in their computing and electronics industries during the 1990s. By 2002, 9 Asia countries (excluding Japan) had a larger share of world exports in ICT than the countries of the EU 15 when intra-EU and external trade are combined. This is only somewhat apparent in the inter-country models. Wixted (2005) has shown through industry-bilateral trade flow maps that the Asian economies transitioned from essentially low technology exporters in 1970 to being specialised in the manufacturing of electronics and computing by 2000. For example, Malaysia’s electronics industry, as Best comments: ‘accounts for half of Malaysia’s total exports and employs a quarter of the manufacturing labor force. The annual rate of growth of manufacturing exports during 1970 to 1995 was over 25 percent per year and electronics accounts for two thirds of manufacturing exports. Over the same 1970 to 1995 period … manufactures share [of exports] increased from 11 percent to 80 percent’ (1999, p. 2).

The process of development is not complete yet, particularly with the rapid rise of China post 2000, but Best (1999) notes that countries which have managed to develop modern electronics industry have also made significant general economic gains, although Malaysia’s per capita income still lags behind others in Asia. The global production network research by Ernst (discussed in Chap. 4) and the production fragmentation approach of Athukorala, for example, (2003) both emphasise that these new production activities have much to do with the relocation of multi-national enterprise operations. Although Ernst (2003) and Chen (2002) emphasise the developing technological capability and national characteristics in Asian countries, they focus on the role of firms, with less information on what features of particular geographies have been influential in attracting the investment. Many accounts of East Asian development have arisen, but amongst the most popular has been the ‘flying geese model’ (e.g. see Blomquist 1997, Barker and Goto 1998 and Mathews 2003). Not only has it been seen as an empirical assessment of history, but a theory which is useful for policy development. In essence, the flying geese conceptualisation of development envisages products within a hierarchy of technological complexity and production costs. As products move up the hierarchy, production centres successively move between the countries of Asia. What complicates this picture is that some components have seemingly never been outsourced from Japan (see Matthews 2003). Undoubtedly, there has been some movement of production as costs rise and technological capabilities improve, however, the flying geese model underplays the interaction between different players in constructing different components of products. Barker and Goto point to a spatial structure in the relevant industries in Asia that has commonalities with what has already been observed for interdependencies in transport, computing and electronic goods across the EU. They comment: ‘production in the Asia-Pacific region is becoming interdependent. It is organised in the form of coincidental hierarchical networks stretching across different countries, driven by the strategies of firms who look to comparative and technological advantage of particular territorial sites for particular parts of the productive processes. Production structures are thus increasingly linked across countries’ (1998, p. 269).

184

9

Cluster Complexes: Electronics and ICT

This indicates a high degree of integration in the structuring of intra-regional and inter-regional networks, that couldn’t be completely captured by the cluster network diagrams. Intra-Asian trade is now greater than intra-European activity in office and telecommunications equipment (Table 9.7). Table 9.7 also indicates that there is more trade between Asian economies than between Asia and all other destinations. This is despite the final markets for the products remaining predominantly in higher income countries, particularly the USA (Athukorala 2003). To complement the analysis already presented, the intra-Asian spatial relations are depicted in Fig. 9.9 based on data in Ng and Yeats (2003). It is based on data Table 9.7 Regional trade – office & telecom equip 2002 ($b & change)

Intra-Asia Intra-Western Europe Asia to North America Asia to Western Europe North America to Asia Latin America to North America

Value

Annual percentage change

2002

1995–00

2001

2002

211.3 169.5 108.4 76.0 47.4 32.4

10% 10% 5% 8% 9% 24%

−14% −7% −21% −16% −20% 0%

14% −5% 4% 1% −7% −6%

Source: WTO (2003) Table IV.41 http://www.wto.org/english/res_e/statis_e/its2003_e/its03_ bysector_e.htm

Fig. 9.9 Major Intra-Asian exports of components 2001 (share of all Asian trade in parts)Source: Based on original data in Ng and Yeats (2003) Notes: Unlike many of the other charts in this book, the arrows indicate the flow of goods. The flows are calculated as a share of all Asian trade in parts and components with a cut-off of 1%

9.4

Contemporary and Emerging Nodes of Global ICT

185

on bilateral trade in parts and components summed across all industries, which as we have seen these countries are highly specialised in computing and electronics. Some of the major features of the East Asian production system evident in Fig. 9.9 are: • Japan is a major destination for intra-regional exports for many East Asian countries; • Japan has a number of significant bilateral relations: Taiwan’s exports to Japan (2.3%) imports from Japan (2.8%) South Korean exports to Japan (1.4%) imports (3.4%) and China’s exports (4.5%) and imports (6.7%); and • Of all links the strongest is China’s exports to Hong Kong (11% of all trade between East Asian 10 economies). This current position is the result of progressive evolution towards more intensive levels of regional integration, particularly intra-regional intra-industry trade (see Blomquist (1997, p. 53). Such patterns and trends in production and trade raise the question of whether the rise of East Asian manufacturing has been based simply on the transfer of overseas technology to where labour is cheap or whether it reflects national and regional sets of capabilities? Unfortunately, constructing a profile of the technological capacity is somewhat more difficult than trade analysis. Understanding indigenous innovative capability is problematic even though East Asia has been one of the key topics in industrialisation and innovation research (see e.g. Cardoza 1999, Saxenian and Hsu 2001, and Chen 2004). Roessner et al., for example, who developed an index of high technology capabilities reveals that data on external patenting by relevant countries has too many uncertainties and so they use ‘Non-Resident Applicant data’ because it provides: ‘the most effective solution in terms of coverage, ease of use, and availability. Although this indicator is the least capable of speaking directly to a nation’s ability to innovate, it does provide insight into other nations perceptions of a nation’s capacity to develop, produce, and market new technology’ (2001, p. 37).

Thus, there are challenges with whatever data is chosen, but Table 9.8 presents direct data from the U.S. Patent Office on foreign applications for patents by Asian countries. Japan and Ireland, together with Australia and Canada for reference purposes, as two medium sized economies. Most Asian countries do not have a strong patent profile in the USA but Taiwan, in particular, and, to a lesser degree, South Korea have both managed to grow their level of patenting. Taiwan has a stronger patenting performance in the USA than the UK, while South Korea although stronger than most, is patenting less than Canada or France. Several factors need to be considered here. First, the Asian economies are focused on the USA market which may provide an incentive to patent in America and therefore inflate the relativities. Second, there are known disincentives for taking out patents on electronics based technologies because the speed of change makes secrecy a better protection (see Hall and Ziedonis 2000 and Graham 2003). However, as this would be the same for all players, the relativities are relevant. Finally, it should be noted that the tables includes all patenting in the

186

9

Cluster Complexes: Electronics and ICT

Table 9.8 Percentage of foreign U.S. patent applications: 1994–99 Residence of inventor

1994

1995

1996

1997

1998

1999

Germany UK Ireland France Finland Canada Australia Japan South Korea Singapore Taiwan Hong Kong China Malaysia Thailand Israel India Brazil Mexico Total foreign United States

13.7 5.9 0.2 5.5 0.9 5.1 1.1 45.7 2.8 0.2 4.3 0.3 0.1 0.0 0.0 1.3 0.1 0.2 0.1 82624.0 107233.0

13.4 5.9 0.1 5.7 0.8 5.4 1.1 45.1 3.2 0.2 4.6 0.2 0.2 0.0 0.0 1.2 0.1 0.1 0.1 88419.0 123958.0

13.1 5.4 0.1 5.1 0.9 5.0 1.2 44.7 4.8 0.2 5.4 0.3 0.2 0.0 0.0 1.2 0.1 0.2 0.1 88295.0 106892.0

13.0 5.4 0.1 5.0 0.9 5.0 1.1 44.1 5.2 0.2 5.8 0.2 0.1 0.1 0.0 1.3 0.1 0.1 0.1 94812.0 120445.0

12.9 5.7 0.2 4.9 0.9 5.3 1.3 42.1 5.1 0.3 6.9 0.3 0.2 0.0 0.0 1.3 0.2 0.2 0.1 107579.0 135483.0

14.1 5.8 0.2 5.2 1.1 5.1 1.2 39.7 4.2 0.4 7.8 0.3 0.2 0.1 0.0 1.7 0.2 0.2 0.1 120362.0 149825.0

Source: National Science Board (2002) Appendix Table 6–13. Note: Blanks = 0.0

USA. Given the technological specialisation of Asia in electronics, it would be expected that most patents from these countries would relate to electronics. This is not the case for the other countries. Below is a sketch of technological capabilities for a few countries in East Asia, developed from a variety of research sources. Taiwanese local firm strengths include: • PC components – motherboards, mouse pointers, monitors and scanners (Matthews 2003:21). • Taiwan is a major DRAM (semiconductor) producer (Ernst 2001). • Taiwanese firms have been able to align local strengths with international production networks - ‘characterised by vertical disintegration but with strong linkages between local firms and across national borders’ (Chen 2002, p. 22). Taiwanese local firm weaknesses include: • Taiwanese firms have been unable to move into hard disk drives, video cassette players and recorders and ‘until the late 1990s’ flat panel displays (Matthews 2003, p.21). South Korean local firm strengths include: • semiconductors (DRAM), advanced computer displays (Ernst 2001, p. 2). • sophisticated mass produced electronics – microwave ovens, TVs and picture tubes (Ernst 2001, p.13).

9.4

Contemporary and Emerging Nodes of Global ICT

187

• South Korean local firm weaknesses include: firms are focussed on price sensitive hardware rather than software (Ernst 2001, p. 2). The local firm strengths of other East Asian countries include: • • • •

Television sets, DRAM semiconductors and PC components (Ernst 2001, p. 7). The local firm weaknesses of other East Asian countries include: Differentiated design based products, services and software (Ernst 2001, p 7). Japanese consumer electronics was initially outsourced to Malaysia and Thailand was then moved to China (Ernst 2001, p. 8). • Malaysia has three specific weaknesses; significant reliance on imports due to restricted local supply chains, export reliance on the USA market and a concentration in low end assembly products (Ernst (2002, p. 41). Looking across all the evidence, it appears that the combination of low cost production and the design features of ICT equipment, specifically the huge variety of modular components which are easily transportable, has enabled the Asian regional production complex specialising in computing and electronics. Traded and knowledge interdependencies have developed spatially within Asia and are linked with the other producers of these products across the world.

9.4.3

Emergent Nodes in the Global Architecture of Production

Beyond the rise of East Asia, two other nodes in the global architecture of production are worth analysing.

9.4.3.1

Ireland

A great success story of ICT-driven economic growth, which was often quoted, particularly in the late 1990s and early 2000s was the Celtic Tiger (Ireland).9 Ireland is considered to have two ICT related clusters – the first is based on foreign direct investment in hardware production facilities, while the second is a local software industry (Roper and Grimes 2003). The following analysis pays particular attention to the former cluster, questioning the degree to which it is a cluster at all under existing uses of that term. The EU 15 international inter-regional model (1995) indicates that Ireland’s office and data processing national-meso cluster was reliant to an extraordinary degree on imported components (61% of output) and from the United Kingdom alone (over 30% of all inputs from a single country – 50% of imports).

9 Irish GDP doubled between the early 1990s and early 2000shttp://www.esri.ie/content.cfm?t = Irish%20Economy&mId = 4 viewed 15 Sept 2004.

188

9

Cluster Complexes: Electronics and ICT

Table 9.9 Import intensity Imports/Value Added (1994–2003) Office, & Computing Machinery Radio, TV & Comms Equip

1994

1995

1996 1997 1998 1999

2000

2001 2002 2003

3.7

3.5

2.4

3.4

5.1

6.0

4.4

5.0

6.1

5.8

5.5

6.5

5.9

4.6

2.1

2.9

2.0

2.7

3.2

5.8

Data Source (OECD 2005b) Note: Due to data issues this should be considered indicative only

A time series of import intensity (Table 9.9) lends plausibility to the result. For office and computing machinery, Ireland’s requirement for imports is quite significant across the period considered. Radio, TV and Communications equipment exhibited a high ratio at the beginning of the period dropped, but early in the 2000s began to rise again. The OECD rated Ireland as having the highest import to export ratio for ICT in OECD countries in both 1995 and 2000 when compared against the import propensity of manufacturing (2002a, p. 32). The OECD also found Ireland to have the largest ICT trade surplus as a percentage of manufacturing trade (2002a, p. 29). When the argument, however, is that this group of activities represents a cluster bounded by the political borders of Ireland, the degree of dependence on imports is highly relevant. The literature on Ireland focuses on the performance of the ICT sector’s growth and as a driver of a modern knowledge based economy. For example, Green et al. write: ‘As an emerging knowledge-based economy, Ireland has one of the highest concentrations of information and communications technology (ICT) activity and employment in the OECD. This activity comprises primarily electronics hardware manufacturing, such as personal computers (PCs); software products and services, especially business application products and “localisation”; and call centres. It has largely been driven by foreign direct investment (FDI), although recent evidence suggests that indigenous firms are now growing at a much faster rate than the multinational sector. It is also a major factor in high-skill, high-wage job creation since the early 1990s and comprises a key element of Ireland’s national innovation system, which is also a regional one in the European context’ (2001, p. 47).

There is little mention of either the scale of imports, or their source(s). Green et al. do, however, acknowledge the important role of payments for technologies. ‘Yet the economy’s dependence on FDI has also encouraged the use of imported rather than locally generated technologies. This is reflected in the very large deficit in Ireland’s “technology balance of payments”, which measures flows in knowledge and “disembodied” technologies between countries’ (2001, p. 50) [emphasis added].

Predominantly, it is the growth and importance of exports (see Green et al. 2001, p. 49) that receives attention. Roper and Grimes (2003) research a range of emerging IT clusters (Helsinki, Tel Aviv and Dublin) but only mention imports in connection with the Finnish IT cluster centred on Helsinki. Without denying that Ireland has made substantial progress in improving per capita GDP and the living standards of its population off the back of rapid industrialisation, O’ Sullivan (1999) is somewhat circumspect about the capability of Irish industry.

9.5

Conclusions and Implications

189

‘we are far from being able to understand the long term significance of the current boom. In part, the problem is that the evidence available is ambiguous. The main issue, however, is a dearth of empirical studies that might allow us to understand the innovative capacity’ (1999, p. 277).

9.4.3.2

China

The speed and scale at which China is emerging appears to be unlike other experiences of economic development in the world. ‘For the last two decades, China’s rise in international trade has been outstanding. China’s share in international trade more than trebled, jumping from less than 2% in 1985 to about 7% in 2005. China has become the third largest exporter in the world in 2004 and is expected to become the first largest by the beginning of the next decade’ (Gaulier et al. 2007, p. 209).

What is obvious from the evidence presented in this chapter and in the emerging literature is that there is an evolving re-orientation of trade linkages within Asia to include China. Evidence in Gaulier et al. (2007) suggests than China in terms of volume is becoming a major actor in the Asian system, in particular re-orienting Japanese trade patterns. However, Tan and Kohr (2006) emphasise that currently China is the low cost producer and assembler in the region. Gaullier et al. (2007) warn that China’s price competitiveness may in the long run be self-harming. On the current evidence, it will take another decade before it becomes clear how the rise of China reconfigures the external trade network as well as generating its own internal structure of clusters. During this time, its internal technological capability should be monitored and not assumed on the basis of technology classifications created for developed economies (noting the cautions of Srholec 2007).

9.5

Conclusions and Implications

Far from diminishing the value of location, the analysis presented in this chapter suggests that a high level of imports is commonplace for computing and electronics production. This is not discussed widely, apart from the structure of Asia’s economies. However, it appears to be a part of the nature of the technology – its modular nature facilitates and encourages both specialisation and high levels of trade. Given this, together with the evidence on the strength of the linkages between a number of countries, for example Ireland and the United Kingdom or some of the Asian economies, it is misleading to analyse clusters in isolation from their network of suppliers and buyers. The reluctance highlighted here, to acknowledge the overall evidence on imports seems to stem from a deeper problem, the view that it would deny evidence on the development of a local skills base and a local supply architecture (promoted by Porter 1990). More importantly, it again points to the absence of framework within

190

9

Cluster Complexes: Electronics and ICT

the systems of innovation research than integrates the emergence of new clusters initially based on FDI and imported components and the continuing economic and innovation strength of existing clusters. Overall, trade networks remain remarkably constant, with new developing production centres fitting into the existing networks, rather than creating something totally new. The one exception to this might be China. The USA, Japan and Germany all emerged as important nodes for supplies. In turn, these primary nodes are themselves being supplied by countries in East Asia. Export patterns10 across Asia are revealed as both strongly intra-regional as well as focussed on Japan and the USA. Interestingly, although East Asian countries have a strong intra-regional trade pattern, they do not seem to have developed a European styled key hub system. Germany, is a dominant supplier of computing and electronics products across the spatial scale of Europe, but is not so central on the global scene. However, within Europe there has been the emergence of the UK and France as second tier suppliers, perhaps indicating the further fragmentation of specialisation. This analysis also begins to suggest the structure of modular value chains (see Sturgeon 2003) as opposed to vertically integrated production (aerospace), develops within a different spatial framework, even though the modelling has been conducted at a highly aggregated level. As computing and electronics products integrate small modular units composed of easily transportable components, emerging IT clusters,11 have been able to specialise in complementary rather than competing products, as Bresnahan et al. indicates: ‘Emerging ICT clusters in Israel, India, Ireland and Taiwan all have significant ties with the United States, which helped them to exploit the ICT-intensive US growth of the second half of the 1990s. In this respect, it was critical for these regions to position themselves in product spaces that were complementary to the main sources of demand (notably Silicon Valley and the US) rather than directly competing with them’ (2001, p. 851).

This finding is of the greatest significance. It connects the local and the global economic environments. It indicates that cluster development is not an isolated creation but occurs in association with the wider technical communities that are embedded in particular clusters. Therefore, policies aiming at developing or encouraging ICT development need to take account of both the global geography of development and the underlying existing and emerging technologies which underpin the fragmentation of production.

10 Notwithstanding Figure 9.9 and the trade specialisation maps in Wixted (2005 Appendix 2), it would have been valuable to have been able to construct better I-O modelling for Asia. 11 Ireland, Cambridge UK, Israel, Scandinavia, India, Taiwan and Northern Virginia.

Chapter 10

Conclusions on the Architecture of Economies

‘For the economic actor, space exists only to the extent that others have given shape to it by their actions. Input-output represents this space by a web of supplier-user relationships. Past economic actions create field forces. Methodological individualism’s atomistic view supposes great degrees of freedom for individuals, and this presupposes a homogeneous space, which amounts to no space at all. In contrast, we find it more realistic to assume that two actors, once they have interacted, can be bound by some relationship and sometimes not even be separable’ DeBresson (1996, pp. 151–152).

10.1

Atolls of Innovation or Something Else?

The neo-Schumpeterian research agenda has a wide variety of foci, but during the 1990s it became increasingly dominated by a geographic emphasis on systems of innovation, which are typically described as national or sub-national (clusters and regions). The attitude of Freeman (2002, p. 209) in choosing to concentrate ‘on developments at the national level in the belief that the major phenomena of forging ahead, catch-up and falling behind in nineteenth and the twentieth centuries can most plausibly be explained in terms of national systems’ appears to be representative of the attitude. The sub-national systems perspective, although focusing on a different spatial scale, also characteristically terminates with national borders. At both the national and sub-nation systems level, the interest is in studying the history of places, looking for endogenous regional factors of success and policy options for the future. As the study of systemic innovation has seemingly taken on this limited spatial framework, the research for this book was designed to investigate the degree to which this claim is justifiable and whether it is an appropriate paradigm for future research. Specifically I wanted to investigate: 1. Does the ‘innovation systems’ agenda give primacy to political geography (nation states and regions within nations) over other possible frameworks including economic space, which may lead to more research on extra-territorial links and if so, why; 2. What is it about linkages between economic actors (relationships particularly between users and producers) within innovation theory that makes them so important for the development and diffusion of new products and services;

B. Wixted, Innovation System Frontiers, Advances in Spatial Science, DOI: 10.1007/978-3-540-92786-0_10, © Springer-Verlag Berlin Heidelberg 2009

191

192

10

Conclusions on the Architecture of Economies

3. What is known about product and knowledge linkages that extend beyond the borders of particular innovation systems (especially industrial clusters); 4. In multi-spatial research of innovation systems is interregional input-output modelling useful and can it reveal linkages between systems that are statistically important; 5. Does the structure of international input-output relations make sense when other evidence on knowledge flows, technological specialisations and national and sub-national systems of innovation are considered; and 6. What new insights into systems of innovation and the interdependencies between them emerge from understanding the spatial structure of linkages? These specific questions can be summarised as whether innovation systems are delineated by borders or whether it is more appropriate to follow production in economic space to combine the advantages of the analysis of location together with trade in a multi-spatial perspective. To follow these lines of connection between places it is necessary to focus on the role, scale and spatial structure of interdependencies that cross borders (intra or inter-nationally). A particular interest, here, was the desire to uncover whether clusters develop within wider networks of clusters or have no particularly discernable extra-territorial structural dimension. To meet the objectives of this research it was determined that being able to analyse a large number of industries and geographic locations was a better approach than a case study of just one cluster (objective four above). Thus, a number of inter-country input-output datasets along with modelling software were constructed for the research. The results have broadly revealed that innovation systems, with few exceptions are perceived to be co-terminus with nation states, even when clusters are studied. They are indeed sub-national systems. Inter-country modelling has proven to be a useful tool. Interdependencies across borders are important in some production systems and the spatial dimension is informative for comprehending innovation system development. Input-output modelling for this purpose, however, has a number of drawbacks which have been described in depth in Chap. 5. Despite these shortcomings, interregional input-output data still presents unique opportunities for the analysis of multi-spatial network structures. Input-output economics has a long tradition (Rose and Miernyk 1989) of developing a range of mathematical tools for measuring the linkages between industries and working on multi-regional analyses (Polenske and Hewings 2004). The modelling conducted for the present research was based on calculations of net output multipliers and thus imports as a share of each dollar of output. This has allowed the results to be easily interpreted and analysed alongside the existing literature on clustering and internationalisation. Against the backdrop that the analysis here has provided an image of the architecture of multi-spatial innovation systems, it is hoped that future research will examine the role of multinational corporations, key universities, government policies, infrastructure such as highways and even physical geography play in forming the shape of inter-cluster networks. It is also hoped that it will be possible to focus on different dimensions of innovation systems (knowledge flows, talent, and services vs. product transfers) rather than the more restricted analysis possible here.

10.1

Atolls of Innovation or Something Else?

10.1.1

193

Local Agglomeration and Specialisation

A vast literature makes it clear that the concentration of related economic activities, horizontally (similar) or vertically (suppliers and buyers) should be understood as a structural dimension of development. Larger agglomerations tend to be centres of industrial production or services (Sassen 2002) as well as the hubs of innovative effort. For example, 21 statistical geographic zones across Europe account for more half of the high technology patents applied for to the European Patent Office (Eurostat 2002). Additionally, multinational firms been slowly internationalising their R&D effort away from traditional centres (OECD 2003b) but the pace of change is increasing (OECD 2008b). This combination of factors has resulted in only a slow change to national technological specialisations (Laursen 1998b). The persistence of locations as places of economic advantage has often been argued to be based in the intensity of interdependencies (buzz) that attract new players and encourage technology adoption and creativity in new product development and the marketing of innovations. The range of specialised products and services, the information transfer embedded in user–producer relations enabling innovation and the local chatter of people in the same industry (untraded knowledge flows) assisting with problem solving are all interdependencies that encourage the co-location of businesses (see Chap. 3). Indeed, long-term user-producer connections are so important for innovation that they can be considered a predictive variable (von Hippel 1988 and DeBresson 1999). Interdependencies, in the view of Edquist (1997b), are one of the dominant themes of systems of innovation research. The concern to study the endogenous features of particular places is understandable. Nonetheless, although national specialisations are relatively stable (Wolff 2000 and Metcalfe et al. 2002), they are not perfectly so. Laursen (1998b) has shown that bilateral trade patterns are slowly de-specialising and the trend in industry location in Europe appears to be favouring dispersion rather than geographic concentration (see Storper et al. 2002). The re-location of production has been observed in low technology industries such as TCF (Gereffi 1998) and in high technology alike (Matthews 2003). Indeed a large part of economic activity, but maybe not knowledge production, in computing and electronics has been accumulating in East Asia (see Chap. 9). With such changes occurring globally, it is no surprise that the processes of locational change can also be observed intra-nationally (Hewings et al. 1998). Looking to the future Pavitt observes that: ‘firms specializing in systems design and integration are not postindustrial. They are instead the prolongation of the industrial system into a period of growing specialization and complexity, and of growing capacities to store, transmit and manipulate information. High wage countries may indeed find themselves specializing increasingly on ‘services’, but not as an alternative to manufacturing activities but as the skill-intensive components within them. The Visible Hand of manufacturing will not become invisible (Langlois 2001), but continue to exploit economies of physical scale, speed, and scope. At the same time, the Visible Brain of systems integration could become the dominant form of business organization in the world’s advanced countries’ (2003b, pp. 18–19).

Pavitt’s perspective is generally reinforced by the research presented here and specifically supported by the data on modular technologies such as electronic components.

194

10

Conclusions on the Architecture of Economies

The research presented in this book also reveals we aren’t there yet, as many industrial economies are still in the game in terms of manufacturing. Nevertheless, it is a more promising framework than that of the jurisdictionally bounded innovation systems approach. Until quite recently, it was difficult to identify a body of literature that specifically drew attention to the limitations of this latter approach. This is beginning to change, see for example Carlsson (2006), Wolfe and Gertler (2004) and Simmie (2004), all of whom draw attention to an urgent need to incorporate extra-regional linkages and not requiring all dimensions of Porter’s diamond to lie inside a region. The present book has attempted to begin this work of defining ‘functional’ innovation systems by revealing the extent of their significant international economic relationships. Rather than affixing interdependencies within geographies, it is necessary to follow them through economic space, which being infrequently continuous or self-sufficient and thus requiring a multi-spatial perspective. What gives a particular location an advantage and how is that location linked to other locations with different technological advantages – in a given production system? There is considerable evidence that multinational enterprises through global production networks and intra-firm trade connect the capabilities of different places to one another. However, while other fields of research (world cities in particular) have been developing an understanding of the spatial linkages and hierarchy of economic geography and activity, neo-Schumpeterians have by and large stuck to comparative analysis of systems. The current research has identified a number of findings on the role, scale and spatial structure of interdependencies that extend across borders.

10.1.2

Bridging Local and Global: The Role of Interdependencies

In the earlier chapters of this book, particularly Chap. 3, the role of interdependencies within local or national economies was discussed at length. A recurring finding is that the processes which drive knowledge (and above all, tacit knowledge) diffusion prefer proximity. For example, technology spillovers via national linkages have a higher correlation with export market shares than spillovers through international imports (see Laursen and Meliciani 2001). It is apparent that there are three types of interdependencies, namely: • Trade – the transfer of goods and services; • Long term business-to-business interactions (user–producer relations); and • The untraded flow of tacit information through local talk. The impact that these types of interdependencies on innovation can be summarised as: • A mechanism for the transmission of economic growth (and declines) between regions; • Generating structural rigidities and therefore promoting less rapid changes in economic structure than might be expected other wise (see discussion in Chap. 4);

10.1

Atolls of Innovation or Something Else?

195

• The acquisition of embodied innovation [access to the benefits of the innovative work of others]; • The development of user–producer relations facilitate the transfer of commercial information and thus encourage innovation, and • Information that assists with solving technical and other context based problems. Some evidence for all three types of connections extending across borders has been identified. Naturally, given the methodologies adopted, the strongest evidence was found for the role of traded goods and services with the implication of embodied technology transfers. The weakest evidence was on the nature of business-to-business relationships across borders. Nevertheless, evidence from the literature on industrial marketing techniques and global production networks, for example, suggests that trans-border associations do transfer knowledge and would therefore be the basis of innovative activity. Interestingly, very strong, albeit indirect, evidence was discovered in a range of literatures on the possible role of person-to person knowledge flows in trans-border networks. The connection between social networks, migration patterns and trade is a new field, but the work of Rauch (1999 and 2001) and others reveal that the movement of people and the movement of goods across the world are closely associated. More direct evidence on untraded interdependencies extending around the world is provided by Saxenian1 and Hsu (2001) who have studied the technical communities in Taiwan and Silicon Valley. These communities work together not just through virtual electronic communications, but also physically, through frequent visits in both directions.

10.1.3

The Scale of Connectedness – Do These Links Matter?

Going from the literature to the research presented here, a significant effort was devoted to measuring the scale of trans-border traded interdependencies. In the first instance, Chap. 6 revealed a range of industries require significant levels of imported inputs. A few of these were resource based industries such as petroleum and a number were, the expected, cost based industries such as textiles, clothing and footwear. Among them, and, and high on the list were industries with higher R&D intensities. Aerospace, electronics and communications, office and computing equipment, and motor vehicles all emerged as dependent to varying degrees on imported components. The pharmaceuticals and instruments industries required noticeably less imported inputs. While the traditional OECD classification of industry by technology was unhelpful in grouping high import-intensive activities, a modified version of Pavitt’s (1984) innovation taxonomy was more useful. The majority of high import industries in both the OECD model and EU model were identified as scale based, with a few

1

More recently she has written of role of tans-border venture capitalists (2006).

196

10

Conclusions on the Architecture of Economies

science-based sectors also included. Analysing this further, some results suggested that technological complexity was a useful concept to introduce into future research on clustering and production geography fragmentation. The findings on the scale of individual linkages between clusters, appears to be generally supportive of the conclusions of other inter-regional I-O studies including those of Hewings et al. arising from a multi-regional analysis of the Chicago metropolitan area. In that study, the authors found that ‘while the interindustry relationship generates circulation of economic activity and hence creates impacts outside the region of original stimulus, the size of these impacts is relatively small’ (2001, p. 214). For the most part the bilateral international inter-cluster linkages are quite small. The top three countries in Europe; Germany, the UK, and France and in the OECD model, Japan and the USA often retained the highest share of value added domestically. Knowing that many I-O studies reveal only small inter-regional flows of value but being aware that studies of international trade have indicated concentrated specialisation patterns it was, nevertheless, a surprise to discover the actual strength of many connections between countries. When pared back to the main trunk routes for trade it became very apparent that the scale of cross-border traded interdependencies were economically significant and they had very definite spatial structures. The research has uncovered a number of bilateral links between countries at the national cluster level that are statistically important. Most countries were dependent on imports from one or two key suppliers and, in some cases, such as Ireland’s reliance on the UK for office and computing components, this dependence was extraordinary high.

10.1.4

Innovation Atolls?

Innovation, then, doesn’t seem to occur simply in a series of hotspots2 like island atolls on an ocean and neither is the world universally similar in innovativeness. The evidence, suggests that industry, place and space all matter and thus there is a need for multi-spatial conceptualisations of innovation. We need a highway map, or better still diagrams of the structural features of the form and functions of geographic locations within the global architecture of production and innovation. As Sturgeon states: ‘We need to better understand the various roles that local agglomerations play within spatially extensive value chains and begin to map the activities that tend to concentrate in particular places even as the geographic ‘footprint’ of linked economic activity expands. It is the linkages mechanisms, between firms and between places that especially deserve more of our research attention’ (2003, p. 200 emphasis added).

Problematically, it is necessary to have regular updates to understand the evolving patterns of development. 2

Such as Florida’s (2005) ‘the world is spiky’ – although it is that.

10.2

Systems of Systems: Sectoral Footprints

10.2 10.2.1

197

Systems of Systems: Sectoral Footprints The Spatial Structure of Clustering and Networking

It is already known that manufacturing goods are generally traded in accordance with economic gravity – the largest nearest market (Wolf 2000) and innovative regions have been found to positively impact on neighbouring regions (Beaudry and Breschi 2003). Business services appear to follow a different pattern with business services linked across distant nodes (e.g. cities, see Beaverstock et al. 1999). A nodal structure to external linkages also appears to be characteristic of some biotechnology clusters (Gertler and Levitte 2003). Chapter 7 (motor vehicles), 8 (aerospace) and 9 (computing and electronics and communications equipment) each explored in detail the spatial structure of intercountry cluster networks. The analysis indicates that each production system has its own individual system of international flows. Further, all countries had primary dependency links. Results from analysing structures between 1970 and 2000 reveal that the patterns of inter-country inter-cluster linkages have been changing and that the approach adopted in the current research project can be used to highlight intertemporal changes (see also Wixted and Cooper 2007). Lastly, these changes in structure often appeared to be new connections bolted onto an existing network framework, although this is not always the case (aerospace). Evidence presented in this book suggests that beyond the locality of individual clusters there are three characteristics of the spatiality of clustering. These are: • Intra-continental macro-regional clustering, e.g. the multi-state clustering of auto production in the U.S and Canada (see e.g. the work of Klier 1999); • Each national cluster had specialised its international linkages forming individual cluster complexes of interdependencies usually having different network partners as suppliers and buyers; and • Groups of cluster complexes can form larger structures in the global economy – called here: exo-nets (e.g. in auto production the clusters of European countries are not significantly linked to the clusters in the Asia–Pacific and Americas). An apparent oddity of economic development is that these findings seem relevant to economic geography at the different scales of micro, city and macro clustering.

10.2.2

Clustering, Fragmentation and Integration – Economic Fractals

Production agglomerates at levels as small as individual streets through to those of national significance, as Malmberg and Maskell point out: ‘One problem relates to the issue of spatial scale. The notions local and regional, which are often central in analyses of spatial clustering, are extremely elastic. First, the two notions are often used more or less as synonyms in the literature. Furthermore, they may denote a

198

10

Conclusions on the Architecture of Economies

number of geographical scales, extending from the local neighbourhood (a street or block in a city, or a small town) through to entire nations or even groups of nations. Similar mechanisms or forces are held to explain both why advertising agencies flock together at a particular street (Madison Avenue) in New York, and why the ‘European banana’ developed as a core area of heavy industrialization during the 19th century, an area which is extended across several countries in the heartland of what is now the European Union. It does not, however, seem possible to define, once and for all, a specific geographical scale at which one could argue that agglomeration economies exert a particularly strong influence. Rather, it seems reasonable to allow the scale to vary according to which type of phenomenon that is emphasized in the analysis’ (2002, p. 442).

It seems natural to allow scale to be scalable, but this commentary misses the interdependencies inherent in the structure. Madison Ave isn’t just an agglomeration of its own; it exists within a global city with a number of other agglomerations of professional firms (legal, financial and others) and their international links. There must be a number of micro cluster complexes within New York. By way of another example, it may be the case that apparel production in Montreal, once co-located is increasingly geographically fragmented along sub-sectoral lines (design vs. production) within the city. At a more aggregated level, Hewings observes of America’s inter-state trade: ‘essentially, establishments in any state are trading more with establishments outsidethe state and less with those within the state; as a result, intrastate multipliers are declining while interstate feedback effects are increasing’ (2007). The conclusion thus comes around again that we need to pay attention to the spatial logic of development (Tellier 1997). Dodgshon (1998) for example, indicates that across large periods of time there have been macro trends in the patterns of geographic development.

10.2.3

Cluster Complexes: Nodes, Flows and Hierarchies

At the macro level the evidence is that manufacturing is not entirely moving to developing economies as suggested by Pavitt (2003a, b), but there is an important interplay between the developing technology regimes and the organisation of production. Nevertheless, the location of centres of design, manufacturing and systems integration is a fundamental issue. The modelling presented here has shown that the scale of cross-border interdependencies is significant and rather than adopting a vague notion of globalisation, connectedness is highly specialised. In doing so, the research results have generated a series of insights into the phenomena of clustering which include the structure, organisation and innovativeness of cluster complexes. The charts of multi-country inter-cluster connections in Chaps. 7, 8 and 9 suggest that there are three structural dimensions of cluster complexes. These are: • Locational nodes – including the local labour market, technological specialisations, industrial specialisations, institutions and policy, • Linkages – weak and strong ties with other places; and • Hierarchy – the strength of individual regions as measured by the number of other regions that link to it for supplies.

10.2

Systems of Systems: Sectoral Footprints

199

There is an extensive literature on the locational features of clusters but the evidence presented here indicates that just as firms form business-to-business networks, clusters also seem to be specialising within networks to locate themselves within international value systems. All national clusters analysed above, have particular linkages that represent a high share of imports and some of these are so strong it would appear to make little sense to think in terms of nation-state clusters. These cluster networks tend to, differing degrees, agglomerate around particular supplier clusters. In the production systems examined, countries such as the UK was the supplier of only a few countries, with France doing only a little better. In all cases, Germany had a clearly economically superior hierarchical position in Europe, with all other countries in the Europe Union 15 highly dependent on its exports. Japan and the USA also emerged as important hub economies in all three sectoral systems either with in the Asia–Pacific–Americas or globally. This result can’t be divorced, unfortunately, from the economic scale of these countries and the limitations of modelling,3 which as already noted, were based on entire economies. The results do, however, appear to reflect the global competitive positions of sub-national systems.

10.2.4

Spatial Organisation: Assembly, Integration and Technologies

Finally, the diagrams of cluster complexes made it possible to visualise the organisation of production in different value systems. For example, the spatial structure of motor vehicle production has some similarities to the structure of aerospace value networks presumably on the basis that they have pooled interdependencies (assembly integration) at the geographic scale. They differ markedly, however, from the extra-territorial linkage structure in computing, electronics and communications equipment – presumably on the basis of the more modular technology and organisational structures in the latter. Although the spatial shape of production differs from product to product, Chaps. 7, 8 and 9 suggest that different spatial structures are related to particular product typologies. The two assembler complexes, or pooled interdependencies in the language of Thompson (1967), of motor vehicles and aerospace consisted of network structures closely associated with first tier nodes. On the other hand, the office and computing and the radio, TV and communications (electronics) cluster complexes spread across a more diverse geographic system. Curiously, although the evidence is incomplete, data presented in Chap. 8 suggests that the aerospace production system is in the process of transforming from being very structured and 3 For example while large economies would be expected to dominate the imports structure of smaller economies, it is unlikely that a small economy will be able to capture 10% of the imports share of larger economies. It would be better if the modeling could be developed at the scale of the individual clusters rather than nation states.

200

10

Conclusions on the Architecture of Economies

hierarchical to one with many active nodes and a spatial structure somewhere in between the auto and ICT systems. The analysis presented here has made it possible to see that motor vehicle cluster complexes were separated into an Asia–Pacific–Americas exo-net and a European exo-net based around purchases from Germany. The data for aerospace revealed that the USA was the major node for supplies across Europe and OECD countries. What these two production systems have in common is that they are both based around large scale assemblers: Such integrators must bring together a wide range of technologies into relatively high priced products mass consumer markets (autos) or very specialised business investments (aerospace).4 In contrast, the ICT cluster networks are much more diversified. Although Germany is the primary node in the EU, the UK and France are both nodes of more importance, than was the case in transport and the UK, France and Germany are more dependent on the USA and Japan than they were in the motor vehicles industry. In turn, export data for East Asian countries suggests that the USA and Japan had become, by the end of the 1990s, heavily dependent on supplies from Taiwan, South Korea and others countries in Asia. This web of cluster complexes would also appear related to the multi-configurational nature of the technology. Not only is there a wide variety of components being assembled into final products, but there is a huge variety of final products and an equally broad price range. Various electronic components are integrated5 into everything from iPods, organisers, and cell phones through to continental wide information and command systems (defence and air traffic control). This variety of component specifications, price, manufacturing quality and position in the technological product life cycle (etc.), all allow clusters in ICT to capture small niches of the market, more than is possible in assembler oriented production systems. However, research on this bundle of topics (organisation, geography and technology) requires much more analysis.6

10.3

Conclusions and Implications

Although, the emphasis throughout has been that the current analysis of systems which is circumscribed by borders is inappropriate, this doesn’t mean that borders are irrelevant. Borders, even within nation states influence the level of trade flows.

4 Brusoni et al. (2001) show that aerospace assemblers need to be knowledgeable about the technologies required for their products even if they don’t build those technologies. 5 Integrators are still strong and need to know more than they make, but decentralised governance of design is prevalent (see Dibiaggio 2007).) 6 It is interesting to note that a recent article by Rosenkopf and Schilling (2007) highlights the complexity of business alliance network patterns for aerospace, motor vehicles and ICT/electronics as well as others such pharmaceuticals.

10.3

Conclusions and Implications

201

Rather than taking borders as a ‘fact’ neo-Schumpeterians could make an invaluable contribution by questioning what it is about borders that influences innovativeness and trade. There are already a few clues in the limited spread of knowledge. This would be in contrast to neo-classical economics which is ‘puzzled’ by the role of borders, particularly across jurisdictions with low tariffs barriers.

10.3.1

Clusters Don’t Innovate in Isolation

This would also continue to move us further away from the neoclassical presentation of technology as an exogenous ‘manna from Heaven’ or knowledge as a freely available good. The neo-Schumpeterian literature has been on a trajectory that is moving away from the ‘Schumpeter Mark I innovator’ (Malerba and Orsenigo 1996), the heroic individual entrepreneur who brings inventions to the marketplace. The whole paradigm shift in innovation studies has been towards considering its systemic nature. At the level of businesses, we know now that ‘entrepreneur[s] cannot innovate alone’ it takes networks of firms to bring innovations to the market (DeBresson 1999). Further still, it has been observed that the combinations of history, institutions, government policies, legal frameworks, education systems and even the competencies of competing businesses, in particular geographic places, influence the innovativeness of businesses. However, to date, these innovation spaces have been mostly understood as limited by national borders, a view that can be contrasted with that of Leamer and Storper (2001), who suggest that the global division of labour is geographically specialising and growing more connected simultaneously. If, as the research presented here suggests, the relationships between clusters closely resembles the dynamics inside clusters, in terms of risk reduction, supplier networks and social relationships, why should innovation systems be co-terminus with political geographies? Why would we not examine the extension of production systems, organizations and the utilisation of competencies across regional or international borders? The analysis presented here supports a view, which is not new, that clusters and national innovation systems are not atolls of innovation but are intricately embedded within the wider architecture of global production. Saxenian and Hsu, for example, state that ‘as governments around the world clamour to establish venture capital industries and technology parks in efforts to replicate the Silicon Valley experience, the Taiwanese case suggests that new centres of technology and entrepreneurship cannot be created in isolation’ (2001, p. 917).

Such a finding should come as no surprise – innovation stems from, as many authors (e.g. Lundvall 1992b) have emphasised previously, a social process of learning, which are by their nature interactive and interdependent. Even clusters, then, seem to rely on interdependencies within systems of systems.

202

10.3.2

10

Conclusions on the Architecture of Economies

Implications for Theory

Even if the evidence presented here on production systems and by implication innovation, is only shadows of the innovative systems as they might be better understood (e.g. the specific geography of knowledge hasn’t been a focus), there are important implications. That clusters don’t innovative in isolation reveals that the old dichotomy between relative proximity and globalisation within the parallel debates over clustering and production fragmentation apparent in the innovation and other literature has serious deficiencies. First, it has been shown that many industries are reliant, to a significant degree, on imported components and intermediate goods. Further, technology oriented industries are fragmenting their production and thus technological specialisation appears to be an important factor in the geographic spread of production. Next, in many cases there is a particular trans-border transaction structure for each particular production system. Therefore, within the research agenda on systems of innovation there needs to be a shift in the intellectual resources away from the investigation of the isolated place alongside knowledge spillovers to examine the role of place in the development of international value systems. Product complexity, modularisation, and the role of corporate networks as an organisational form to marshal the different forms of capital (physical, technological and human, etc.) all need specific investigation. Most importantly, there is an urgent need for a general effort to develop systems of innovation approaches that can account for the development and integration of new nodes (spatial analysis) within international production. Which nodes have been successful in becoming ‘joined up’ to wider networks, which have failed and why? Unlike the production fragmentation research, the analysis presented here has revealed that resource processing, assembly, and the production of multi-configurable technologies are most likely to become increasingly fragmented.7 This suggests that the lowering of transport and communication costs, the commonly noted dynamic behind fragmentation (See e.g. Hewings 2007) may be incomplete.8 Complexity (see Simon 1962) of the production-process is probably an important variable.

10.3.3

Implications for Policy

This book has not been focussed on assessing the policy implications of the emerging structure of international business and innovation, instead focussing on the empirical

7 Both modularity and the types of assembly encompassed here both require high degrees of architectural control (see Rosenkopf and Schilling 2007) 8 It is worth noting that OECD argues that the relative cost of transporting goods has not been reduced so ‘the impact of distance on the structure of trade has not declined over time’ (2008a, p. 106).

10.3

Conclusions and Implications

203

and theoretical issues. This was a careful choice because policy is often in the specifics rather than general frameworks and because in opening up this new branch of research it was essential to establish some language and empirical guideposts. However, it is worth making a few notes on policy, in passing here. A few of the big themes of cluster policy development do seem to have been largely agreed. For example Andersson et al., comment: ‘Among the various approaches available, broker policies should aim at strengthening the framework for dialogue and cooperation by the various relevant stakeholders involved in clusters, and not favour individual players. Demand side policies should seek to increase openness to new ideas and innovative solutions. Training policies may be targeted at upgrading skills and competencies which are essential for effective clustering of SMEs. Measures for the promotion of international linkages should be designed with a view to enhancing the interplay between foreign and domestic actors. Framework policies, finally, should put in place an over-riding playing field marked by effective and consistent rules for inter-actor transactions. Both hard-defined aspects such as social capital and attitudes, and habits that support trust in transactions should be taken seriously by policymakers’ (2004, p. 27).

Cortright (2006, p. 48) takes a largely similar perspective on policy development. He suggests that policies aiming at enhancing the ‘micro-foundations of industry clustering’ should be build around: • • • • • • •

‘Labor market pooling: labor market information, specialized training; Supplier specialization: brokering, recruiting, entrepreneurship, credit; Knowledge-spillovers: networking, public sector research and development support; Entrepreneurship: assistance for startups, spin-offs; Lock-in: work to extend, refine, and recombine existing distinctive specializations; Culture: acknowledge and support cluster organization; and Local demand: aggregate and strengthen local demand’.

Unsurprisingly, these policy options reflect the centrality of territories to the interests of politicians and researchers. With the hindsight of the analysis presented in the present work, the aforementioned suggestions for cluster policies can be seen to avoid the value chain setting in which clusters exist, the hierarchical position of existing activities, the opportunities for change or the opportunity to access global networks. All are important issues, and all avoided in above suggestions. Cluster policies are no panacea and neither is moving into high technology fields. Most importantly, cluster policies would appear to make most sense if intimately connected to trade, technology and value chain management policies and programs. Curiously, in the light of the evidence presented here, place-based policies particularly those focussed upon skills, seem to offer some opportunities for developing economies. The cluster complexes model opens up the opportunity for such countries to design policies that aim at complementary skills, technologies and production activities rather than directly competing. In acknowledging the systemic properties of innovative clusters to enter global value chains and production networks, developing countries need to concentrate on building groups of firms and supporting institutions for developing the appropriate labour pool.

204

10.3.4

10

Conclusions on the Architecture of Economies

Future Research

The research reported here has only been the beginning. It has only sketched new possibilities for research frontiers. There is more left unexamined than has been investigated. Many areas need further research, some of them have been indicated already, but a few of the important possibilities include: 1. At the level of organisations; • The organisation and management of product complexity and modularisation (the economics and engineering principles that make it viable to break some products down into smaller and smaller modules along with the geographic fragmentation of production); • The network dimension – what are the strategies of multinational corporations with regard to managing both the geography and product modularisation of their global production networks; 2. At the level of technologies; • The co-evolution of technological strengths and global product hierarchies, the contribution of local knowledge to global products, and the ongoing evolution of trade structures; and 3. At the spatial level; • The role of borders in shaping innovation and trade; • Examining the hypothesis that cluster complexes will exist at spatial scales other than across national borders; • Examining the apparent seeming regularities in the patterns of clustering and fragmentation across cities, regions and nation states. This is a vast canvas, but given the past successes of the neo-Schumpeterians, we can look forward to new findings that continue to extend and challenge our understanding of social and economic processes.

References

AAAS (2004a) Trends in NIH R&D, Request vs Actual [1978–2005] www.aaas.org/spp/rd/ guihist.htm AAAS (2004b) Trends in AAAS Report XXIX: Research and Development FY 2005. www.aaas. org/spp/rd/guihist.htm Acha V, Cusmano L (2005) Governance and co-ordination of distributed innovation processes: patterns of R&D co-operation in the upstream petroleum industry. Econ Innovat New Tech 14(1–2):1–21 AECMA [European Association of Aerospace Industries] (2002a) European Aerospace Industry Industrial and structural Issues. ACEMA, Brussels AECMA [European Association of Aerospace Industries] (2002b) The European Aerospace Industry Facts and Figures 2001. AECMA, Brussels AFFA [Agriculture, Fisheries and Forestry – Australia] (2001) Australian Food Statistics 2001. AFFA, Canberra AFFA [Agriculture, Fisheries and Forestry – Australia] (2002) Australian Food Statistics 2002. AFFA, Canberra Airbus Industrie (2004) A380 Programme Update January 2004 http://www.airbus.com/media/ programme.asp viewed February 2004 (no longer on website) Airbus Industrie (2007) A380 Navigator http://events.airbus.com/A380/default1.aspx accessed 16 November 2007 Amendola G, Guerrieri P, Padoan PC (1998) International patterns of technological accumulation and trade. In: Archibugi D, Michie J (eds) Trade, growth and technical change. Cambridge University Press, Cambridge, pp 141–167 Anas A, Arnott R, Small KA (1997) Urban Spatial Structure. The University of California Transportation Center Working Paper March 1997 UCTC No. 357. University of California at Berkeley Andaluz J, Lanaspa L, Sanz F (2002) Geographical dynamics: a sectoral comparison between the economic landscapes of the United States and Europe. Reg Stud 36(4):321–332 Anderson J, O’Dowd L (1999) Borders, border regions and territoriality: contradictory meanings, changing significance. Reg Stud 33(7):593–604 Anderson JE, van Wincoop E (2001) Gravity with gravitas: a solution to the border puzzle. NBER Working Paper Series No 8079, NBER, Cambridge Andersson T, Fredriksson T (2000) Distinction between intermediate and finished products in intra-firm trade. Int J Ind Organ 18:773–792 Andersson T, Serger S, Sörvik J, Hansson E (2004) The cluster policies whitebook. International organisation for knowledge economy and enterprise development (IKED), Malmo Ando M (2006) Fragmentation and vertical inter-industry trade in East Asia North American. J Econ Fin 17:257–281 Archibugi D, Iammarino S (1999) The policy implications of the globalisation of innovation. Res Pol 28(2–3):317–336

205

206

References

Archibugi D, Michie J (eds) (1997) Technology, globalisation and economic performance. Cambridge University Press, Cambridge Archibugi D, Michie J (eds) (1998) Trade growth and technical change. Cambridge University Press, Cambridge Archibugi D, Pianta M (1992) The technological specialization of advanced countries. A Report to the EEC on international science and technology activities. Kluwer, Boston Armstrong H (1995) Convergence among regions of the European Union, 1950–1990. Paper Reg Sci 74(2):143–152 Arndt S (1998) Super-specialization and the gains from trade. Contemporary Economic Policy XVI (October 1998):480–485 Athukorala P-C (2003) Product fragmentation and trade patterns in East Asia. Research school of Pacific and Asian studies working paper No. 2003/21. Canberra, Division of Economics – Australian National University Audretsch D, Feldman M (1996) R&D spillovers and the Geography of innovation and production. Am Econ Rev 86(3):630–640 Audretsch D, Fritsch M (2002) Growth regimes over time and space. Reg Stud 36(2):113–124 Australian Industry Group (2002) Research and Development Expenditure and Drivers in Australian Manufacturing. AiG, Melbourne Baark E (2001) The Dynamics of Innovation in Engineering Consulting Services. Presented to the ECIS conference: The Future of Innovation Studies 21–23 September 2001 Eindhoven, The Netherlands Baldwin C (2007) Frameworks for Thinking About Modularity. Sloan Industry Studies Annual Conference April 25–27 2007 Cambridge Massachusetts http://web.mit.edu/sis07/www/ options-3.htm accessed 29 November 2007 Baldwin C, Clark K (2000) Design rules: Volume 1. The power of modularity. MIT, Cambridge Baptista R (2000) Do innovations diffuse faster within geographical clusters? Int J Ind Organ 18:515–535 Barker B, Goto A (1998) Technological systems, innovation and transfers. In: Thompson G (ed) Economic dynamism in the Asia–Pacific. Routledge, New York Baumol W (2000) What Marshall didn’t know: on the twentieth Century’s contributions to Economics. Q J Econ CXV (1):1–44 Beaudry C, Breschi S (2003) Are firms in clusters really more innovative? Econ Innovat New Tech 12(4):325–342 Beaverstock JV, Smith RG, Taylor PJ (1999) A roster of World Cities. Cities 16(6):445–458 Begg I, Fagerberg J, Guerrieri P. (1999) Conclusions and policy implications. In: Fagerberg J, Guerrieri P, Verspagen B (eds) The economic challenge for europe: adapting to innovation based growth. Edward Elgar, Cheltenham, pp 228–237 Bergman EM, Charles D, den Hertog P (2001) In Pursuit of Innovative Clusters. In: OECD Innovative clusters: drivers of national innovation systems. OECD, Paris, pp 7–15 Bernard M, Ravenhill J (1995) Beyond product cycles and flying geese: regionalization, hierarchy and the industrialization of East Asia. World Polit 47(2):171–209 Best M (1999) Cluster dynamics in theory and practice: Singapore/Johor and Penang Electronics. Center for Industrial Competitiveness. University of Massachusetts Lowell http://www.uml. edu/centers/CIC/publications.htm Blomquist H (1997) Economic interdependence and development in East Asia. Praeger, Connecticut Botham R, Simpson E, Whyte V (2001) Scottish Aerospace: an Actual or Embryonic Cluster? unpublished report for Scottish Enterprise, from the Training and Employment Research Unit, University of Glasgow: www.scottish-enterprise.com/about/what/research/ Breschi S, Malerba F (2001) The Geography of innovation and Economic clustering: some introductory notes. Ind Corp Change 10(4):817–833 Breschi S (2000) The Geography of innovation: a cross-sector analysis. Reg Stud 34(3):213–229 Breschi S, Lissoni F (2001a) Knowledge spillovers and local innovation systems: a critical survey. Ind Corp Change 10(4):975–1005

References

207

Breschi S, Lissoni F (2001b) Localised knowledge spillovers vs innovative Milieux: knowledge ‘Tacitness’ reconsidered. Pap Reg Sci 80(3):255–273 Bresnahan T, Gambardella A, Saxenian A (2001) Old economy inputs for the “New Economy” outcomes: cluster formation in the New Silicon valleys. Ind Corp Change 10(4):835–860 Brusoni S, Prencipe A, Pavitt K (2001) Knowledge specialisation, organisational coupling, and the boundaries of the firm: why do firms know more than they make? Admin Sci Quart 26(4):597–621 Bucken-Knapp G (2001) Just a train-ride away, but still worlds apart: prospects for the øresund region as a binational city. GeoJournal 54:51–60 Bunnell T, Coe N (2001) Spaces and scales of innovation. Prog Hum Geogr 25(4):569–589 Capello R (1999) Spatial transfer of knowledge in high technology Milieux: learning versus collective learning processes. Reg Stud 33(4):353–365 Cappelen A, Fagerberg J, Verspagen B (1999) Lack of regional convergence. In: Fagerberg J, Guerrieri P, Verspagen B (1999) The economic challenge for europe: adopting to innovation based growth. Edward Elgar, Aldershot, pp 130–148 Cappelen A, Fagerberg J, Verspagen B (2000) Regional convergence, clustering and European policies presented at the 8th Joseph A. Schumpeter Conference, 28th June – 1 July 2000, University of Manchester Capron H, Cincera M, van Pottlesberghe B (2000) The national innovation system and its international linkages. In: Capron H, Meeusen W (eds) The Belgian national innovation system. Springer, Berlin, pp 117–136 Caracostas P, Soete L (1997) The building of cross-border institutions in Europe: towards a European system of innovation. In: Edquist C (ed) Systems of innovation: technologies, institutions and organisations. Pinter, London, pp 395–419 Cardoza G (1999) Learning and innovation paths in East Asia. Sci Publ Policy 26(4):259–276 Carlsson B (2006) Internationalization of innovation systems: a survey of the literature. Res Policy 35:56–67 Carlsson B, Jacobsson S, Holmén M, Rickne A (2002) Innovation systems: analytical and methodological issues. Res Policy 31(2):233–245 Casellas A, Galley C (1999) Regional definitions in the european union: a question of disparities? Reg Stud 33(6):551–558 Chen S-H (2002) Global production networks and information technology: the case of Taiwan. EastWest Centre Working Papers: Economic Series No. 46. May 2002, East-West Centre, Honolulu Chen S-H (2004) Taiwanese IT firms’ Offshore R&D in China and the connection with the Global innovation network. Res Policy 33(2):337–349 Christensen C, Craig T, Hart S (2001) The Great disruption. Foreign Affairs March/April:80–95 Church A, Reid P (1999) Cross-border co-operation, institutionalization and political space across the English channel. Reg Stud 33(7):643–655 Cizmeci D (2005) An examination of Boeing’s supply chain management practices within the context of the global aerospace industry. Unpublished Master of Engineering in Logistics at the Massachusetts Institute of Technology, Boston, June 2005 Cohen S (2002) Mapping Asian integration: transnational transactions in the Pacific Rim. Am Asian Rev XX 20(3):1–29 Cohen WM, Levinthal DA (1990) Absorptive capacity: a new perspective on learning and innovation. Admin Sci Quart 35:128–152 Combes P-P, Overman HG (2003) The spatial distribution of economic activities in the European Union. Centre for Economic Performance Discussion Papers No. 587. London School of Economics and Political Science, London Combes P-P, Lafourcade M, Mayer T (2003) Can Business and Social Networks Explain the Border Effect Puzzle? CEPII Working Papers No. 2003 – 02, Paris, Centre d’Etudes Prospectives et d’Informations Internationales (CEPII) Cooke P (2001) Regional innovation systems, clusters, and the knowledge economy. Ind Corp Change 10(4):945–974 Cooke P, Morgan M (1998) The associational economy – firms, regions, and innovation. Oxford University Press, Oxford

208

References

Cooke P, Heidenreich M, Braczyk H-J (eds) (2004) Regional innovation systems: the role of governance in a globalized World, 2nd edn. Routledge, New York Cooper RJ (2000) An extension of the block spatial path approach to analysis of the influence of intra and interregional trade on multiplier effects in general multiregional input-output models. In: Reggiani A (ed) Spatial economic science: new frontiers in theory and methodology. Springer, Berlin, pp 303–327 Cooper RJ, Wixted B (2003) Commentary on the Draft Report: The economics and social impacts of the Royal Australian Air Force Base Richmond on the Hawkesbury and surrounding region (15 August 2002. Prepared for the Hawkesbury City Council) Cova B, Salle R (2000) Rituals in managing extra-business relationships in international project marketing: a conceptual framework. Int Bus Rev 9:669–685 Crafts N (1999) Economic growth in the twentieth century. Oxford Rev Econ Policy 15(4):18–34 Creedy S (2002) Boeing Flexes local HdH arm. The Australian, April 12, 2002, (Friday) Features, p 25 Creedy S (2003) Ace fixes sights on Hawker. The Australian, June 13, 2003 (Friday) Finance-Column – Aviation, p 23 Cutler D, Neely J (2007) ‘Factbox – Sizing up the Airbus A380 Double-Decker’http://www.reuters. com/article/companyNewsAndPR/idUSL1267687320071112?pageNumber = 2&sp = true accessed 13 November 2007 Czamanski D, Czamanski S (1977) Industrial complexes: their typology structure and relation to economic development. Pap Reg Sci 38(1):93–111 Czamanski S (1971) Some empirical evidence of the strengths of linkages between groups of related industries in urban-regional complexes. Pap Reg Sci 27(1):136–150 Dalum B (1992) Export specialisation, structural competitiveness and national systems of innovation. In: Lundvall B-A (ed) National systems of innovation: towards a theory of innovation and interactive learning, Pinter, London, pp 191–225 Davies S, Rondi L, Sembenelli A (2001) European integration and the changing structure of EU manufacturing 1987–1993. Ind Corp Change 10(1):37–75 Davis D, Reeve T (1997) Human capital, unemployment and relative wages in a global economy. Eur Econ Rev 42(2):1613–1633 Davis DR (2000) Understanding international trade patterns: advances of the 1990s. Mimeo available at http://www.columbia.edu/~drd28/Research.htm downloaded 25 August 2004 Davis DR, Weinstein DE (1996) Does economic geography matter for international specialization. NBER Working Paper No. 5706, NBER, Cambridge MA De la Mothe J, Paquet G (1996) Evolution and Inter-creation: the government–business-society nexus. In: De la Mothe J, Paquet G (eds) Evolutionary economics and the new international political economy. Pinter, London, pp 9–34 DeBresson C (1999) An entrepreneur cannot innovate alone: networks of enterprises are required. Presented at the DRUID Summer Conference 9–12 June 1999 www.druid.dk DeBresson C (ed) (1996) Economic interdependence and innovative activity an input-output analysis. Edward Elgar, Cheltenham DeBresson C, Amesse F (1991) Networks of innovators: a review and introduction to the issue. Res Policy 20(5):363–379 DeBresson C, Murray B (1984) Innovation in Canada: a report to the Science Council of Canada, 2 vols. CRUST reprint, New Westminster DeBresson C, Murray B, Brodeur L (1986) L’Innovation au Québec. Les Publications du Québec, Quebec DeBresson C, Sirilli G, Hu X (1996) The structure of innovative interactions in the Italian Economy 1981–1985. In: DeBresson C (ed) Economic interdependence and innovative activity: an input output analysis. Edward Elgar, Cheltenham, pp 104–119 DeBresson C, Hu X, Drejer I, Lundvall B-A (1998) Innovative activity in the learning economy: a comparison of systems in 10 O.E.C.D. Countries. OECD, Paris Department of Industry, Technology and Commerce (1987) Measures of science and innovation: australian science and technology indicators report. AGPS, Canberra

References

209

Diamond J (1998) Guns, germs and steel: a short history of everybody for the Last 13,000 Years. Vintage, London Dibiaggio L (2007) Design complexity, vertical disintegration and knowledge organization in the semiconductor industry. Ind Corp Change 16(2):239–267 Dicken P, Thrift N (1992) The organization of production and the production of organization: why business enterprises matter in the study of geographical industrialization. Trans Inst Br Geogr 17(3):279–291 Dicken P (2003a) Global production networks in Europe and East Asia: the automobile components industries. Global Production Networks Working Paper No. 7, Manchester Business School – School of Geography, Manchester Dicken P (2003b) Global shift: reshaping the global economic map in the 21st century, 4th edn. The Guilford, London Dicken P (2004) Geographers and “globalization” (yet) another missed boat? Trans Inst Br Geogr 29:5–26 Diehl M (2001) International trade in intermediate inputs: the case of the automobile industry. Kiel Working Paper No. 1027, Kiel Institute of World Economics, Kiel Dietzenbacher E, van der Linden JA (1997) Sectoral and spatial linkages in the EC production structure. J Reg Sci 37(2):235–257 Dietzenbacher E, Romero I (2007) Production chains in an interregional framework: identification by means of average propagation lengths. Int Reg Sci Rev 30(4):362–383 DITR (2003) Partnering in the global context: an Australian aerospace industry for the 21st century. Report of the Aerospace Industry Action Agenda, Department of Industry, Tourism and Resources, Canberra, http://www.industry.gov.au/assets/documents/itrinternet/AIAA_ RepFin__20050606094337.pdf accessed 14 November 2007 Dodgshon R (1998) Society in time and space: a geographical perspective on change. Cambridge University Press, Cambridge Doloreux D (2004) Regional innovation systems in Canada: a comparative study. Reg Stud 38(5):481–494 Dosi G (1988) Sources, procedures and microeconomic effects of innovation. J Econ Lit 26:1120–70 Dosi G (1999) Some notes on national innovation systems and production, and their implications for economic analysis. In: Archibugi D, Howells J, Michie J (eds) Innovation policy in a global economy. Cambridge University Press, Cambridge UK, pp 35–48 Dosi G, Castaldi C (2002) Local and divergent patterns of technological learning within (partly) globalized markets. LEM Working Paper 2002–22. Pisa Sant’Anna School of Advanced Studies, Pisa Dosi G, Pavitt K, Soete L (1990) The economics of technical change and international trade. Wheatsheaf, Brighton Dosi G, Freeman C, Fabiani S (1994) The process of economic development: introducing some stylized facts and theories on technologies, firms and institutions. Ind Corp Change 3:1–45 Dosi G, Orsenigo L, Labini M (2002) Technology and economy. Laboratory of Economics and Management [LEM] Working Paper Series 2002/18, Pisa Sant’ Anna School of Advanced Studies, Pisa Dowrick S, DeLong B (2001) Globalisation and convergence. Presented at the NBER conference Globalization in Historical Perspective, 4–5 May 2001. Santa Barbara, California Drejer I (1999) Technological change and interindustrial linkages: introducing knowledge flows in input-output studies. Dissertation, Aalborg Drejer I (2000) Comparing patterns of industrial interdependence in national systems of innovation – a study of Germany, the United Kingdom, Japan and the United States. Econ Syst RESearch 12(3):377–399 Drejer I, Kristensen FS, Laursen K (1999) Studies of clusters as a basis for industrial and technology policy in the Danish economy. In: OECD (ed) Boosting innovation: the cluster approach. OECD, Paris

210

References

Drysdale P, Garnaut R (1994) Trade intensities and the analysis of bilateral trade flows in a many country world: a survey. In: Garnaut R, Drysdale P (eds) Asia Pacific regionalism. Readings in International Economic Relations, HaperEducational, Sydney Düring A, Schnabl H (2000) Imputed interindustry technology flows – a comparative SMFA analysis. Econ Syst Res 12(3):363–375 Dyer J, Singh H (1998) The relational view: cooperative strategy and sources of interorganizational competitive advantage. Acad Manage Rev 23(4):660–679 Dyer J, Nobeoka K (2000) Creating and managing a high performance knowledge-sharing network: the Toyota case. Strategic Manage J 21:345–367 The Economist (1992a) Economic growth: explaining the mystery. The Economist, January 4th 1992, pp 17–20 The Economist (1992b) Innovation: the machinery of growth. The Economist, January 11th 1992, pp 19–21 The Economist (1996a) The mystery of growth. The Economist, May 25th 1996, pp 15–16 The Economist (1996b) Economic growth: the poor and the rich. The Economist, May 25th 1996, pp 23–25 The Economist (1998) The strange life of low-tech America. The Economist, October 17th 1998, pp 81–82 The Economist (2003a) Extinction of the car giants. The Economist, June 14th–20th 2003, p 11 The Economist (2003b) One hell of a birthday, Bill. The Economist, June 14th–20th 2003, pp 57–60 The Economist (2003c) When small is beautiful: how big should a nation-state be? The Economist, 20 December 2003, p 103 The Economist (2004) Emerging economies: what’s in a name? The Economist 17 January 2004, p 11 Edquist C (ed) (1997a) Systems of innovation: technologies, institutions and organisations. Pinter, London Edquist C (1997b) Systems of innovation approaches – their emergence and characteristics. In: Edquist C (ed) Systems of innovation: technologies, institutions and organisations. Pinter, London, pp 1–35 Edquist C (2001) The systems of innovation approach and innovation policy: an account of the state of the art. Presented at the DRUID Summer Conference, 12–15 June 2001, Aalborg Denmark Egger P, Pfaffermayr M, Wolfmayr-Schnitzer Y (2001) The international fragmentation of value added: the effects of outsourcing to eastern Europe on productivity, employment and wages in Austrian manufacturing. Osterreichisches Instit Fur Wirtschaftsforschung, Vienna (WIFO http://www.wifo.ac.at/) Elak GG, Jalowiecki B (2002) European boundaries: unity or division of the continent? Reg Stud 36(4):409–419 Elam M (1997) National imaginations and systems of innovation. In: Edquist C (ed) Systems of innovation: technologies, institutions and organizations. Pinter, London, pp 157–173 Ernst D (2000) Carriers of cross-border knowledge diffusion: information technology and global production networks. Paper presented at the 8th Joseph A. Schumpeter Society Conference, 28 June – 1 July 2000, Manchester Ernst D (2001) Moving beyond the commodity trap? trade adjustment and industrial upgrading in east Asia’s electronics industry. East West Centre Economics Series working papers No 10. East West Centre, Honolulu Ernst D (2002) Global production networks in east asia’s electronics industry and upgrading perspectives in Malaysia. East West Centre Economics Series working papers No 44, East West Centre, Honolulu Ernst D (2003) Pathways to innovation in the global network economy: Asian upgrading strategies in the electronics industry. East West Centre Economics Series working papers No 58, East West Centre, Honolulu Ernst D, Kim L (2002) Global production networks, knowledge diffusion, and local capability formation. Res Policy 31(8–9):1417–1429

References

211

Ernst D, Ozawa T (2002) National sovereign economy, global market economy and transnational corporate economy. East West Centre Economic Series working papers No 42, East West Centre, Honolulu European Commission (2002) Innovation policy in Europe 2002: European trend chart on innovation. Office for Official Publications of the European Communities, Luxembourg European Commission (2003) Third European report on science and technology report 2003: towards a knowledge-based economy. European Commission, Luxembourg European Information Technology Observatory [EITO] (2004) EITO 2004 Yearbook. Frankfurt, EITO. www.eito.org Tables and Figures: EITO 2004 – ICT markets incl, UK Eurostat (2000) Harmonised input-output tables 1995. Eurostat, Brussels Eurostat (2002) Research and development in the regions of the EU: regional R&D expenditure highly concentrated. More than half of all High-Tech Patent Applications Submitted by 13 Regions out of 211 in 2000. Eurostat News Release No 31/2002 - 13 March 2002, Eurostat, Brussels http://epp.eurostat.cec.eu.int/portal/page?_pageid = 1090,1137397&_dad = portal&_ schema = PORTAL Evangelista R, Iammarino S, Mastrostefano V, Silvani A (2002) Looking for regional systems of innovation: evidence from the Italian innovation survey. Reg Stud 36(2):173–186 Fagerberg J (1996) Technology and competiveness. Oxford Rev Econ Policy 12(3):39–51 Fagerberg J (1998) User-producer interaction, learning and comparative advantage. In: Archibigi D, Michie J (eds) Trade growth and technical change. Cambridge University Press, Cambridge, UK Fagerberg J (2001) Technology and the wealth of Nations. A critical essay on economic growth theory. Presented at The Future of Innovation Studies, 20–23 September 2001, Eindhoven Falconbridge JR (2007) London’s and New York’s advertising and law clusters and their networks of learning: relational analyses with a politics of scale? Urban Stud 44(9):1635–1656 Feenstra R (1998) Integration of trade and disintegration of production in the global economy. J Econ Persp 12(4):31–50 Feser EJ, Bergman EM (2000) National industry cluster templates: a framework for applied regional cluster analysis. Reg Stud 34(1):1–19 Feser E, Sweeney S, Renski H (2005) A descriptive analysis of discrete U.S. industrial complexes. J Reg Sci 45(2):395–419 Fields G (2006) Innovation, time, and territory: space and the business organization of Dell computer. Econ Geogr 82(2):119–146 Fitzgibbons S, Holmes J, Kumar P, Rutherford T (2003) The Ontario automotive parts industry. Presented to Innovations Systems Research Network Annual Meeting, Ottawa, April 30 – May 2, 2003 Florida R (2005) The world is spiky: globalization has changed the economic playing field, but hasn’t leveled it. Atlantic Month 296(3):48–51 Fontagne L, Freudenberg M (1997) Intra-industry trade: methodological issues reconsidered. CEPII Working Papers No 97–01, CEPII, Paris Freeman C (1974) The economics of industrial innovation, 1st edn. Penguin, Hammonsworth Freeman C (1987) Technology policy and economic performance: lessons from Japan. Frances Pinter, London Freeman C (1994) The economics of technical change. Cambridge J Econ 18(5):463–514 Freeman C (1995) The ‘national system of innovation’ in historical perspective. Cambridge J Econ 19(1):5–24 Freeman C (2002) Continental, national and sub-national innovation systems – complementarity and economic growth. Res Policy 31(2):191–211 Frenken K (2000) A complexity approach to innovation networks: the case of the aircraft industry. Res Policy 29(2):257–272 Frenken K, Leydesdorff L (2000) Scaling trajectories in civil aircraft (1913–1997). Res Policy 29(3):331–348 Friedman T (2000) The Lexus and the Olive tree. New York, Anchor Books Fujimoto T (1999) Toyota motor manufacturing Australia in 1995: an emergent global strategy. Actes du GERPISA No 26–3

212

References

Furman JL, Porter ME, Stern S (2002) The determinants of national innovative capacity. Res Policy 31(6):899–933 Garreau J (1981) The nine nations of North America. Boston, Houghton Mifflin Garrett-Jones S, Wixted B, Turpin T (2003) Mozambique: science, technology and knowledge indicators, 2003. Report to the Ministério de Ensino Superior, Ciência e Tecnologia (MESCT), [Ministry for Higher Education, Science and Technology] Maputo, Mozambique Gassman O, von Zedtwitz M (1998) Organization of industrial R&D on a global scale. R&D Manage 28(3):147–161 Gaulier G, Lemoine F, Ünal-Kesenci D (2007) China’s emergence and the re-organisation of trade flows in Asia. China Econ Rev 18(3):209–243 Gereffi G (1998) More than the market. More than the state: global commodity chains and industrial upgrading in east Asia. In: Chan S, Clark C, Lam D (eds) Beyond the developmental state. St. Martin’s, New York, pp 38–59 Gereffi G, Sturgeon T (2004) Globalization, employment, and economic development: a briefing paper. MIT Industrial Performance Centre Working Paper Series, MIT-IPC-04–006, June 2004 http://web.mit.edu/ipc/www/ Gertler MS, Levitte Y (2003) Local nodes in global networks: the geography of knowledge flows in biotechnology innovation. Presented to DRUID Summer Conference, “Creating, Sharing and Transferring Knowledge: The Role of Geography, Institutions, Organizations”, 12–14 June 2003, Copenhagen, http://www.druid.dk/ Giuliani E, Rabellotti R, Dijk MP van (eds) (2005) Clusters facing competition: the importance of external linkages. Ashgate, Aldershot Godbout T (1993) Employment change and sector distribution in 10 Countries 1970–1990. Monthly Labor Rev October:3–20 Golich V (1992) From competition to collaboration: the challenge of commercial-class aircraft manufacturing. Int Org 46(4):899–934 Gordon R (2000) Does the ‘new economy’ measure up to the great inventions of the past. J Econ Persp 14(4):49–74 Gorzelak G, Jalowiecki B (2002) European boundaries: unity or division of the continent? Reg Stud 36(4):409–419 Graham S (2003) Hiding in the patent’s shadow: firms’ use of secrecy to capture value from new discoveries. Presented at DRUID Summer Conference 2003 on “Creating, Sharing and Transferring Knowledge”, 12–14 June 2003, Copenhagen, http://www.druid.dk/ Green R, Cunningham J, Duggan I, Giblin M, Moroney M, Smyth L (2001) The boundaryless cluster: information and communications technology in Ireland. In: OECD (2001) Innovative clusters: drivers of national innovation systems. OECD, Paris, pp. 47–64 Gregersen B, Johnson B (1997) Learning economies, innovation systems and European integration. Reg Stud 31(5):479–490 Gregory RG (1993) The Australian innovation system. In: Nelson RR (ed) National innovation systems: a comparative analysis. Oxford University Press, New York Grotz R, Braun B (1997) Territorial or trans-territorial networking: spatial aspects of technology-oriented cooperation within the German mechanical engineering industry. Reg Stud 31(6):545–557 Guerrieri P (1999) Patterns of national specialisation in the global competitive environment. In: Archibugi D, Howells J, Michie J (eds) Innovation policy in a global economy. Cambridge University Press, Cambridge, UK, pp 139–159 Guerrieri P, Milana C (1998) High-technology industries and international competition. In: Archibugi D, Michie J (eds) Trade, growth and technical change. Cambridge University Press, Cambridge, UK, pp 188–207 Guerrieri P, Pietrobelli C (2000) Models of industrial districts’ evolution and changes in technological regimes. Presented at DRUID Summer Conference 15–17 June 2000, Rebild Denmark. www.druid.dk Guo D, Hewings GJD, Sonis M (2003) Temporal changes in the structure of chicago’s economy: 1980 – 2000. REAL Disc Pap Tech Ser 03-T-32, Urbana-Champaign

References

213

Gurría A (2008) The global economy and OECD: distilling lessons from a financial crisis http:// www.oecd.org/document/12/0,3343,en_2649_34487_41420876_1_1_1_1,00.html accessed 20 October 2008 Hall P (1998) Cities in civilisation: culture, innovation and urban order. Weidenfold and Nicolson, London Hall B, Ziedonis R (2000) The patent paradox revisited: an empirical study of patenting in the US semiconductor industry, 1979–95. Rand J Econ 32(1):101–128 Haltiwanger J (2000) Aggregate growth: what have we learned from microeconomic evidence? OECD Economics Department Working Papers No. 267, OECD, Paris Harvey M, McMeekin A, Randles S, Southerton D, Tether B, Warde A. (2001) between demand and consumption: a framework for research. CRIC Discussion Paper No 40. Centre for Research on Innovation and Competition, Manchester Hassink R, Dankbaar B, Corves F (1994) Cross border technology networks in border regions. The Case of of the Euroregion Mass-Rhine. MERIT Research Memorandum 94–018, Maastricht Hatzichronoglou T (1999) The globalisation of industry in the OECD countries. OECD STI Working Papers 1999/2, OECD, Paris www.oecd.org Hauknes J (1999) Norwegian input-output clusters and innovation patterns. In: OECD Boosting innovation: the cluster approach. OECD, Paris, pp 61–90 Hayward K (1987) Airbus: twenty years of European collaboration. Int Affairs 64(1):11–26 Head K, Ries J (2001) Overseas investment and exports. Rev Int Econ 9(1):108–122 Heath R (1999) The Ottawa high-tech cluster: policy or luck? In: OECD Boosting Innovation: the cluster approach. OECD, Paris, pp 175–192 Helliwell JF (1995) Do national borders matter for Quebec’s Trade? NBER Working Paper Series No. 5215, NBER, Cambridge Henderson J, Dicken P, Hess M, Coe N, Wai-Chung Yeung H (2001) Global production networks and the analysis of economic development. GPN Working Paper No 1. Manchester Business School – School of Geography, Manchester http://www.art.man.ac.uk/Geog/gpn/index.html Hertog P, den Maltha S (1999) The emerging information and communication cluster in the Netherlands. In: OECD Boosting Innovation: The cluster approach.OECD, Paris, pp 193–218 Henry N, Pinch S (2001) Neo-Marshallian nodes, institutional thickness, and Britain’s “motor sport valley”: thick or thin? Environ Plan A 33(7):1169–1183 Hesseldahl A (2007) Taking the iPhone apart: an analysis from teardown firm portelligent estimates that the new smartphone costs apple a mere $220 to make. BusinessWeek July 2, 2007 http://businessweek. com/print/technology/content/jul2007/tc2007072_957316.htm viewed 13 August 2007 Hewings GJD (2007) On some Conundra in regional science. Ann Reg Sci online 25 October 2007 DOI 10.1007/s00168–007–0160–4 Hewings GJD, Nazara S, Dridi C (2004) Channels of synthesis forty years on: integrated analysis of spatial economic systems. J Geogr Syst 6(1):7–25 Hewings GJD, Okuyama Y, Sonis M (2001) Economic interdependence within the Chicago metropolitan area: a Miyazawa analysis. J Reg Sci 41(2):195–217 Hewings GJD, Sonis M, Guo J, Israilevich PR, Schindler GR (1998) The hollowing out process in the Chicago economy 1975–2015. Geogr Anal 30:217–233 Hicks D, Breitzman T, Olivastro D, Hamilton K (2001) The changing composition of innovative activity in the US a portrait based on patent analysis. Res Policy 30(4):681–703 Hill Senator the Hon. Robert (2003) Australia wins more JSF work. Minister of Defence Press Rellease 162/2003. http://www.minister.defence.gov.au/Hilltpl.cfm?CurrentId = 3384 viewed 1 September 2004 Hoen A (2002a) Identifying linkages with a cluster-based methodology. Econ Syst Res 14(2):131–146 Hoen A (2002b) An input-output analysis of European integration. North-Holland, Amsterdam vom Hofe R, Chen K (2006) Whither or not industrial cluster: conclusions or confusions. Ind Geogr 4:2–28 Hofer R, Polt W (1998) Evolutionary innovation theory and innovation policy: an overview. In: Bryant K, Wells A (eds) A new economic paradigm? Innovation-Based Evolutionary Systems, Department of Industry, Science and Resources, Canberra

214

References

Holbrook A, Wolfe D (eds) (2000) Innovation, institutions and territory: regional innovation systems in Canada. Queen’s School of Policy Studies and McGill-Queen’s University Press, Kingston Holbrook A, Wolfe D (eds) (2002) Knowledge, clusters and regional innovation: economic development in Canada. Queen’s School of Policy Studies and McGill-Queen’s University Press, Kingston Holmes S (2006) The 787 encounters turbulence. BusinessWeek June 19 2006 pp 38–40 Howell LJ, Hsu JC (2002) Globalization within the auto industry: Res Technol Manage 45(4):43–49 Hudson R (2002) Changing industrial production systems and regional development in the New Europe. Sussex European Institute Working Paper No. 45/02, University of Sussex http:// www.one-europe.ac.uk Hummels D, Rapoport D, Yi K (1998) Vertical specialisation and the changing nature of world trade. Econ Policy Rev (Federal Reserve Bank New York) June:79–99 Hulsink W, Manuel D, Bouwman H (2007) Clustering in ICT: from route 128 to Silicon Valley, from DEC to Google, from Hardware to Content. Erasmus Research Institute of Management (ERIM) Report Series Research Series in Management IMF [International Monetary Fund] (2000) World Economic Outlook Database http://www.imf. org/external/pubs/ft/weo/2000/01/data/index.htm IMF [International Monetary Fund] (2001) Debt relief for poverty reduction: the role of the enhanced HIPC initiative. August 2, 2001 http://www.imf.org/external/pubs/ft/exrp/debt/eng/ index.htm viewed on 10 August 2004 Industry Canada (1998) The 1998 Canadian biotechnology strategy: an ongoing renewal process. Industry Canada, Ottawa Industry Commission (1995) Research and development. (Report No. 44). AGPS Canberra Isard W, Schooler EW (1959) Industrial complex analysis, agglomeration economies, and regional development. J Reg Sci 1(2):19–33 Ivarsson I (1999) Competitive industry clusters and inward TNC investments: the case of Sweden. Reg Stud 33(1):37–49 Jackson RW (2004) The impacts of Walter Isard on geography. J Geogr Syst 6(1):71–77 Johnston R (1998) The changing nature and forms of knowledge: a review. Evaluation and Investigations Programme 98/16. Department of Employment, Education, Training and Youth Affairs, Canberra Jones R, Kierzkowski H (1990) The role of services in production and international trade: a theoretical framework. In Jones R, Krueger A (eds) The political economy of international trade: Festschrift in Honor of Robert Baldwin. Basil Blackwell, Oxford, pp 31–48 Jones RW, Kierzkowski H (2001) A framework for fragmentation. In: Arndt S, Kierzkowski H (eds) Fragmentation and international trade. Oxford University Press, Oxford, pp 17–34 Jorgenson DW, Ho M, Stiroh K (2003) Growth of US industries and investments in information technology and higher education. Econ Syst Res 15(3):279–325 Katsikeas C, Kaleka A (1999) Import motivation in manufacturer–overseas distributor relationships guidelines for U.S. industrial exporters. Ind Market Manage 28:613–625 Keeble D, Wilkinson F (1999) Collective learning and knowledge development in the evolution of regional clusters of high technology SMEs in Europe. Reg Stud 33(4):295–303 Kitson M, Martin R, Tyler P (2004) regional competitiveness: an elusive yet key concept? Reg Stud 38(9):991–999 Klepper S (2001) The evolution of the U.S. automotive industry and Detroit as its capital. Presented at the DRUID Winter Conference 17–19 January 2002, Aalborg Denmark www.druid.dk Klier T (1999) Agglomeration in the U.S. Auto supplier industry. Econ Persp 23(1):18–34 Knowledge@Wharton (2008) Behind the curve: have U.S. automakers built the wrong vehicles at the wrong time – again? http://knowledge.wharton.upenn.edu/article.cfm?articleid = 2012 published 09 July 2008 accessed 20 October 2008 Kraetke S (1999) Regional integration or fragmentation? The German – Polish border region in a new Europe. Reg Stud 33(7):631–641

References

215

Krugman P (1991) Geography and trade. MIT University Press, Massachusetts Krugman P (1998) What’s new about the new economic geography? Oxford Rev Econ Policy 14(2):7–17 Krugman P, Obstfeld M (1997) International economics: theory and policy, 4th edn. Addison Wesley, New York Landefeld JS, McCulla SH (1999) Wassily W. Leontief, 1906–99: Wassily Leontief and his contributions to economic accounting. survey of current business, March 1999, US Bureau of Economic Analysis, http://www.bea.doc.gov/bea/ARTICLES/NATIONAL/Inputout/ 1999/0399leon.pdf Landes D (1999) The wealth and poverty of nations. Abacus, London Laursen K (1998a) Do export and technological specialisation patterns co-evolve in terms of convergence or divergence? Evidence from 19 OECD Countries, 1971–1991. DRUID Working Papers No. 1998–18. Aalborg Laursen K (1998b) A neo-schumpeterian perspective on the determinants and effects of international specialisation. Dissertation, Aalborg University, Denmark Laursen K, Meliciani V (2002) The relative importance of international vis-á-vis national technological spillovers for market share dynamics. Ind Corp Change 11(4):875–894 Laursen K, Salter A (2001) The fruits of intellectual production: economic and scientific specialisation of OECD countries. Presented to The Future of Innovation Studies Conference 20–23 September 2001, Eindhoven Centre of Innovation Studies, Eindhoven Le Gallo J, Dall’erba S (2003) Spatial econometric analysis of the evolution of the European regional convergence process, 1980–1999. IERSO (IFReDE-GRES) Université MontesquieuBordeaux IV Leamer E, Storper M (2001) The economic geography of the internet age. J Int Bus Stud 32(4):641–665 Leamer EE, Levinsohn J (1995) International trade theory: the evidence. In: Grossman G, Rogoff K (eds) Handbook of international economics, vol 3. Elsevier Science, Amsterdam, pp 1339–1394 Leoncini R, Montresor S (1999) The automobile technological systems. An empirical analysis of four European countries. Presented at EUNIP International Conference 9–10 December 1999, Dublin, Ireland Linden JA van der (1998) Interdependence and specialisation in the European union: intercountry input-output analysis and economic integration. Labyrint, The Netherlands Linden JA van der, Oosterhaven J (1995) European community intercountry input-output relations: construction method and main results for 1965–1985. Econ Syst Res 7(3):249–269 Linden JA, van der Oosterhaven J, Cuello FA, Hewings GJD, Sonis M (2000) Fields of influence of productivity change in eu intercountry input-output tables, 1970–1980. Environ Plan A 32:1287–1305 Lundvall BA, Johnson B, Andersen E, Dalum B (2002) National systems of production, innovation and competence building. Res Policy 31(2):213–231 Lundvall B-A (ed) (1992a) National systems of innovation: towards a theory of innovation and interactive learning. Pinter, London Lundvall B-A (1992b) Introduction. In: Lundvall B-A (ed) National systems of innovation. Pinter, London Lundvall B-A, Borrás S (1997) The globalising learning economy: implications for innovation policy. DG XII, Commission of the European Union, Brussels Lundvall B-A (1996) National systems of innovation and input-output analysis. In: Debresson C (ed) Economic interdependence and innovative activity an input-output analysis. Edward Elgar, Cheltenham Lundvall B-A (1998) Why study national systems and national styles of innovation. Technol Anal Strat Manage 10(4):407–421 Lundvall B-A, Johnson B, Andersen ES, Dalum B (2001) National systems of production, innovation and competence building. Presented to Nelson and Winter DRUID Summer Conference 12–15 June 2001, Aalborg www.druid.dk

216

References

Lundvall B-A, Maskell P (2000) Nation states and economic development: from national systems of production to national systems of knowledge creation and learning. In Clark GL, Feldman M, Gertler M (eds) The Oxford handbook of economic geography. Oxford University Press, Oxford Lung Y (2003) The changing geography of the European automobile system. Cahiers du GRES No. 2003 – 10. Université Montesquieu Bordeaux IV E3i, IFReDE-GRES et GERPISA http:// www.gres-so.org Macpherson A, Pritchard D (2003) The international decentralisation of US commercial aircraft production: implications for US employment and trade. Futures 35:221–238 Mair A, Florida R, Kenney M (1988) The new geography of automobile production: Japanese transplants in north America. Econ Geogr 64(4):352–373 Malerba F (2002) Sectoral systems of innovation and production Res Policy 31(2):247–264 Malerba F, Orsenigo L (1996) Schumpeterian patterns of innovation are technology specific. Res Policy 25(3):451–78 Malmberg A, Maskell P (2001) the elusive concept of localization economies: towards a knowledge-based theory of spatial clustering. Environ Plan A 34(3):429–449 Mankiw NG (1995) The growth of nations. Brookings Pap Econ Activ 1:275–310 Marceau J (1999) The disappearing trick: clusters in the Australian economy. OECD Boosting Innovation: the cluster approach, OECD, Paris, pp 155–174 Marceau J, Wixted B (2002) Innovative sector, innovative city, innovative region? An analysis of innovation at state level in Australia. Presented at the International Sociological Association XV World Congress of Sociology: The Social World in the 21st Century – Ambivalent Legacies and Rising Challenges, 7–13 July 2002, Brisbane Marceau J, Wixted B, Basri E (2001) The Australian processed food product-service system. Commonwealth Department of Agriculture, Fisheries and Forestry, Australia, Canberra Markusen A (1996) Sticky places in slippery space: a typology of industrial districts. Econ Geogr 72(3):293–313 Markusen A (1999) Fuzzy concepts, scanty evidence, policy distance: the case for rigour and policy relevance in critical regional studies. Reg Stud 33(9):869–884 Marshall A (1890/1961) Principles of economics, 9th [Variorum] edn. MacMillian, Cambridge Marsili O, Verspagen B (2002) Technology and the dynamics of industrial structures: an empirical mapping of Dutch manufacturing. J Ind Corp Change 11(4):791–815 Martin R, Sunley P (2003) Deconstructing clusters: chaotic concept or policy Panacea. J Econ Geogr 3:5–35 Masanauskas J (2003a) Deal to help build super jumbos. Herald Sun (Melbourne) February 11, 2003, (Tuesday) News, Pg. 11 Masanauskas J (2003b) Hopes high for jobs on newest jet. Herald Sun (Melbourne) July 3, 2003 (Thursday) News, Pg. 22 Maskell P (2001) Towards a knowledge-based theory of the geographical cluster. Ind Corp Change 10(4):921–943 Maskell P, Lorenzen M (2004) The cluster as market organization. Urban Stud 41(5/6):991–1009 Matthews J (2002) Introduction. Ind Innovation 9:1–5 Matthews J (2003) Catch-up strategies through technological exchange and cooperation: building on the East Asian experience. Invited paper Democratic Pacific Assembly 2003, 20 September 2003, Taipei, Taiwan Maxcy G (1965) The motor industry. In: Hunter A (ed) The economics of Australian industry. Melbourne University Press, Carlton Victoria, pp 494–538 McCallum J (1995) National borders matter: Canada – U.S. regional trade patterns. Am Econ Rev 85(3):615–623 McQuire S (2007) The United States, Japan and the aerospace industry: from capture to competitor? Pacif Rev 20(3):329–350 Meng B, Sato H, Nakamura J, Okamoto N, Kuwamori H, Inomata S (2006) Interindustrial structure in Asia-Pacific region: growth and integration, by using 2000 AIO table. IDE Discussion Paper No. 50. Institute of Developing Economies-JETHRO, Japan

References

217

Metcalfe S (1995) The economic foundations of technology policy: equilibrium and evolutionary perspectives. In: Stoneman P (ed) Handbook of the economics of innovation and technological change. Blackwell, Oxford, pp 409–512 Metcalfe S, Foster J, Ramlogan R (2002) Adaptive economic growth. CRIC Discussion Paper. No 59. Centre for Research on Innovation and Competition, Manchester Morgan K (1997) The learning region: institutions, innovation and regional renewal. Reg Stud 31(5):491–503 Morgan K (2001) The exaggerated death of geography: localised learning, innovation and uneven development. Presented at The Future of Innovation Studies 20–23 September 2001, Eindhoven Centre for Innovation Studies. Eindhoven Munroe DK, Hewings GJD (2000) The role of intraindustry trade in interregional trade in the midwest of the US. presented at the Thirteenth International Conference on Input-Output Techniques, 17–25 August 2000, Macerata National Science Board (2002) Science and engineering indicators 2002. National Science Foundation, Virginia Nazara S, Hewings GJD, Sonis M (2001) Hierarchical spatial interaction: an exploratory analysis. Regional Economics Applications Laboratory Technical Papers No 01-T-12, Urbana-Champaign Nelson RR (1990) Capitalism as an engine of progress. Res Policy 19(3):193–214 Nelson RR (ed) (1993a) National innovation systems: a comparative analysis. Oxford University Press, New York Nelson RR (1993b) A retrospective. In: Nelson RR (ed) National innovation systems: a comparative analysis. Oxford University Press, New York, pp 505–524 Ng F, Yeats AJ (1999) Production sharing in East Asia: who does what for whom, and why? World Bank Working Papers No 2197, World Bank, Washington DC Ng F, Yeats AJ (1999) Major trade trends in East Asia: what are their implications for regional cooperation and growth? World Bank Policy Research Working Paper 3084, World Bank, Washington D.C Nietzsche FW (1888) Gotzen-Dammerung [The Twilight of the Idols]. Maxims and Missiles No 26 Niosi J, Bellon B (1994) The global interdependence of national innovation systems: evidence, limits, and implications. Technol Soc 16(2):173–197 Niosi J, Zhegu M (2002) Montreal aerospace cluster: attractors and dynamics. Presentation to the meeting of the Innovation System Research Network 9–10 May 2002, Quebec City Nolan N, Zhang J, Liu C (2007) The global business revolution, the cascade effect, and the challenge for firms from developing countries. Cambridge J Econ doi: 10.1093/cje/bem016 O’Sullivan M (1999) Industrial development in Ireland: a new beginning? Regional adjustment strategies to technological change in the context of European integration (RASTEI). Working Paper 00–17. Oxford University School of Geography and the Environment, Oxford O’Sullivan M (2000) The sustainability of industrial development in Ireland. Reg Stud 34(3):277–290 Obstfeld M, Rogoff K (2000) The six major puzzles in international macroeconomics: is there a common cause? In: Bernanke B, Rogoff K (eds) NBER macroeconomics Annual 2000, MIT, Cambridge MA OECD (1992a) Technology and the economy, The key relationships. OECD, Paris OECD (1995) Input-output database. OECD, Paris OECD (1996a) The knowledge-based economy. OCDE/GD(96)102, OECD, Paris OECD (1996b) Technology and industrial performance. OECD, Paris OECD (1997) Bilateral trade database. OECD, Paris OECD (1998a) Internationalisation of industrial R&D: patterns and trends. OECD, Paris OECD (1998b) STAN [Structural Analysis] database. OECD, Paris OECD (1998c) Bilateral trade database 1998. OECD, Paris OECD (1999a) Proceedings of the OECD ministerial conference a borderless world: realising the potential of global electronic commerce. Ottawa 7–9 October 1998, OECD, Paris (SG/ EC(99)1/Final) OECD (1999b) OECD science, technology and industry scoreboard 1999: benchmarking knowledge-based economies. OECD, Paris

218

References

OECD (1999c) Managing national innovation systems. OECD, Paris OECD (1999d) Boosting innovation: the cluster approach. OECD, Paris OECD (1999e) OECD historical statistics 1960–1997. OECD, Paris OECD (2000) STAN [Structural Analysis] database. OECD, Paris OECD (2001a) OECD science, technology and industry scoreboard: towards a knowledge-based economy. OECD, Paris OECD (2001b) The new economy beyond the hype. OECD, Paris OECD (2001c) Innovative clusters: drivers of national innovation systems. OECD, Paris OECD (2002a) Measuring the information economy. OECD, Paris OECD (2002b) Bilateral trade database 2002. OECD, Paris OECD (2002c) STAN database (France and the USA). OECD, Paris http://puck.sourceoecd.org/ vl = 1619249/cl = 29/nw = 1/rpsv/statistic/s23_about.htm?jnlissn = 16081307 OECD (2002d) STI Review 27: Special Issue on New Science and Technology Indicators. OECD, Paris OECD (2002e) OECD Input-output tables. OECD, Paris OECD (2003a) The sources of economic growth in OECD countries. OECD, Paris OECD (2003b) Science, technology and industry scoreboard 2003. OECD, Paris OECD (2003c) STAN database (Canada, Germany and UK). OECD, Parishttp://puck.sourceoecd. org/vl = 1619249/cl = 29/nw = 1/rpsv/statistic/s23_about.htm?jnlissn = 16081307 OECD (2004) OECD in figures 2004 edition: statistics on member countries. OECD, Paris OECD (2004b) A new world map in textiles and clothing: adjusting to change. OECD, Paris OECD (2005) Science, technology and industry scoreboard 2005. OECD, Paris OECD (2005b) STAN database. OECD, Paris OECD (2006) OECD input-output tables. OECD, Paris OECD (2007a) OECD reviews of regional innovation. competitive regional clusters: national policy approaches. OECD, Paris OECD (2007b) Staying competitive in the global economy: moving up the value chain. OECD, Paris OECD (2008a) Economic policy reforms: going for growth 2008. OECD, Paris OECD (2008b) The internationalisation of business R&D: evidence, impacts and implications. OECD, Paris Ohmae K (1995) The end of the nation state – the rise of regional economies. Free, New York Oinas P, Malecki E (2002) The evolution of technologies in time and space: from national and regional to spatial innovation systems. Int Reg Sci Rev 25(1):102–131 Olienyk J, Carbaugh R (1999) Competition in the world jetliner industry. Challenges 42(4):60–81 Oosterhaven J (1995) Changing specialisation and interdependency of EC-Economies 1959–1975. Austral J Reg Stud 1(1):51–64 Ozman M (2007) Breadth and depth of main technology fields: an empirical investigation using patent data. Science and Technology Policies Research Center Working Paper Series 07/01, Middle East Technical University, Ankara Paasi A (1999) Boundaries as social practice and discourse: the Finnish – Russian border. Reg Stud 33(7):669–680 Paci R, Usai S (1999) Externalities, knowledge spillovers and the spatial distribution of innovation. GeoJournal 49(4):381–390 Paci R, Usai S (2000) Technological enclaves and industrial districts: an analysis of the regional distribution of innovative activity in Europe. Reg Stud 34(2):97–114 Paoli M, Guercini S (1997) R&D internationalisation in the strategic behaviour of the firm. SPRU STEEP Discussion Paper No 39, Science Policy Research Unit, Sussex Papaconstantinou G, Sakurai N, Wyckoff A (1996) Embodied technology diffusion: an empirical analysis for 10 OECD countries. OECD STI Working Papers 1996(1), OECD, Paris Park B-G (2003) Politics of scale and the globalization of the South Korean automobile industry. Econ Geogr 79(2):173–194 Patel P (1997) Localised production of technology for global markets. In: Archibugi D, Michie J (1997) Technology, globalisation and economic performance. Cambridge University Press, Cambridge, pp 198–214

References

219

Patel P, Pavitt K (1991) Large firms in the production of the world’s technology: an important case of non-globalisation. J Int Bus Stud 22:1–21 Patel P, Pavitt K (1994) The nature and economic importance of national systems of innovation. STI [Science Technology and Industry] Rev 14, OECD, Paris Patel P, Pavitt K (1997) The technological competencies of the world’ s largest firms: complex and path-dependent, but not much variety. Res Policy 26(2):141–156 Patel P, Pavitt K (1998) Uneven (and divergent) technological accumulation amongst advances countries: a framework of explanation. In: Archibugi D, Michie J (eds) Trade growth and technical change. Cambridge University Press, Cambridge, pp 55–82 Patel P, Pavitt K (2000) National systems of innovation under strain: the internalisation of corporate R&D. In: Barrell R, Mason G, O’Mahony M (eds) Productivity, innovation and economic performance. Cambridge University Press, Cambridge, pp 217–235 Paterson M (2000) Car culture and global environmental politics. Rev Int Stud 26(2):253–270 Pavitt K (1984) Sectoral patterns of technical change: towards a taxonomy and a theory. Res Policy 13(6):343–373 Pavitt K (2001) Can the large Penrosian firm cope with the dynamics of technology. SPRU Electronic Working Paper Series No 68, July 2003, SPRU, Sussex Pavitt K (2002) Systems integrators as “post-industrial” firms? Presented at the DRUID Summer Conference on ‘Industrial Dynamics of the New and Old Economy – Who Embraces Whom?’ 6–8 June 2002, Copenhagen, Denmark Pavitt K (2003a) What are advances in knowledge doing to the large industrial firm in the “new economy”? SPRU Electronic Working Paper Series No 91, SPRU, Sussex Pavitt K (2003b) Specialisation and systems integration: where manufacture and services still meet. SPRU Electronic Working Paper Series No 95, SPRU, Sussex Polenske KR (1995) Leontief’s spatial economic analyses. Struct Change Econ Dyn 6:309–318 Polenske KR, Hewings GJD (2004) Trade and spatial economic interdependence. Pap Reg Sci 83(1):269–289 Poon JPH, Thompson ER, Kelly PF (2000) Myth of the triad? Geography of trade and investment blocs. Trans Inst Br Geogr 25:427–44 Porter ME (1990) The competitive advantage of nations. MacMillian, New York Porter ME (1998) Clusters and the new economics of competition. Harvard Bus Rev Nov–Dec:77–90 Porter ME (2003) The economic performance of regions. Reg Stud 37(6–7):549–578 Pred A (1977) City systems in advanced economies. Hutchinson, London Prencipe A (1997) Technological competencies and product’s evolutionary dynamics: a case study from the aero-engine industry. Res Policy 25:1261–1276 Pressi B, Solimene L (2003) The dynamics of clusters and innovation: beyond systems and networks. Physica, New York Pries L (2003) Emerging production systems in the transnationalisation of german car manufacturers: adaption, application or innovation? New Technol Work Employment 18(2):82–100 Pritchard D, MacPherson A (2004) Industrial subsidies and the politics of world trade: the case of the Boeing 7e7. Ind Geogr 1(2):57–73 Rangan S, Lawrence R (1999) A prism on globalization: corporate responses to the dollar. Brookings Institution, Washington Rauch JE (1999) Networks versus markets in international trade. J Int Econ 48:7–35 Rauch JE (2001) Business and social networks in international trade. J Econ 39:1177–1203 Rey SJ (2004) Walter Isard’s influence on analytical human geographic research. J Geogr Syst 6(1):3–6 Rhode P (2001) The evolution of california manufacturing. Public Policy Institute of California, San Francisco Riemens W (2002) Structural changes in the automotive industry in australia: a review of user-producer relationships. Int J Automotive Technol Manage 2(1):101–125 Ritter T (2000) A framework for analyzing interconnectedness of relationships. Ind Market Manage 29(4):317–326

220

References

Roelandt T (1998) Cluster analysis and the systems of innovation approach. research strategies and policy implications. Presented at OECD conference Innovation Systems – Growth Engines for the 21st Century, Sydney 19–20 November 1998 Roelandt T, Hertog P den (1999) Cluster analysis and cluster-based policy making in OECD countries: an introduction to the theme. In: OECD boosting innovation: the cluster approach. OECD, Paris, pp 9–26 Roessner D, Porter A, Jin X-Y, Newman N, Yglesias E (2001) Indicators of technology-based competitiveness: incorporating recent changes in the concept, “high-technology,” and in data availability. Georgia Institute of Technology, Atlanta Rolls Royce (2007) Annual report 2006. Rolls-Royce plc, BuckinghamGate Romer P (1986) Increasing returns and long run growth. J Polit Econ 94:1002–1037 Roper S, Grimes S (2003) Wireless Valley, Silicon Wadi and Digital Island: Helsinki, Tel Aviv and Dublin in the ICT Boom. European Regional Science Association 2003 Congress 27–30 August 2003, Jyväskylä, Finland Rose A, Miernyk W (1989) Input-output analysis: the first fifty years. Econ Syst Res 1(2):229–271 Rosenkopf L, Schilling M (2007) Comparing alliance network structure across industries: observations and explanations. Strategic Entrepreneurship J 1:191–209 Rowthorn R, Kozul-Wright R (1998) Globalization and economic convergence: an assessment. UNCTAD Discussion Papers No. 131. http://www.unctad.org/ Royall I (2001) Boeing coup. Herald Sun (Melbourne) August 22, 2001, Wednesday Business, p 34 Rugman A, D’Cruz J (1993) The double diamond model of international competitiveness: the Canadian experience. Manage Int Rev 1993(2):17–39 Rutherford T, Holmes J (2008) ‘the flea on the tail of the dog’: power in global production networks and the restructuring of Canadian automotive clusters. J Econ Geogr 8:519–544 Saddler D (1999) Internationalization and specialization in the european automotive components sector: implications for the hollowing-out thesis. Reg Stud 33(2):109–119 Sako M (2003) Modularity and outsourcing: the nature of co-evolution of product architecture and organization architecture in the global automotive industry. In Prencipe A, Davies A, Hobday M (eds) The business of systems integration. Oxford University Press, Oxford, pp 229–253 Salter A, D’Este P, Martin B, Geuna A, Scott A, Pavitt K, Patel P, Nightingale P (2000) Talent, not technology: the impact of publicly funded research on innovation in the UK. SPRU, University of Sussex, Sussex Sandven T, Smith K (1998) Understanding R&D intensity indicators – effects of differences in industrial structure and country size. IDEA Paper Series No 14. STEP Group, Oslo Sassen S (2002) Introduction: locating cities on global circuits. In Sassen S (ed) Global networks, linked cities. Routledge, New York, pp 1–36 Saxenian A (1994) Regional advantage: culture and competition in Silicon Valley and Route 128. Harvard University Press, Cambridge, Massachusetts Saxenian A, Hsu J-Y (2001) The Silicon Valley – Hsinchu connection: technical communities and industrial upgrading. Ind Corp Change 10(4):893–920 Saxenian A (2006) The new Argonauts: regional advantage in a global economy. Harvard University Press, MA, Cambridge Scarpetta S, Bassanini A, Pilat D, Schreyer P (2000) Economic growth in the OECD area: recent trends at the aggregate and sectoral level. OECD Economics Department Working Papers No 248, OECD, Paris Schoenberger E (1990) U.S. manufacturing investments in western europe: markets, corporate strategy, and the competitive environment. Ann Assoc Am Geogr 80(3):379–393 Schoemaker P (1990) Strategy, complexity and economic rent. Manage Sci 16(10):1178–1192 Schumpeter J (1968) The theory of economic development: an inquiry into profits, capital, credit, interest and the business cycle. Harvard University Press (first published in English 1934), Cambridge, Massachusetts Science and Innovation Mapping Taskforce (2003) Mapping Australian science and innovation: main report. Australian Government, Canberra

References

221

Scott J (1999) European and North American contexts for cross-border regionalism. Reg Stud 33(7):605–617 Scott R (2002) Flying high in new industry. The Age (Melbourne) May 15, 2002 (Wednesday) Education Section, p 16 Sheehan P, Tikhomirova G (1996) Diverse paths to industrial development in East Asia and ASEAN. CSES Working Paper No 6, Centre for Strategic Economic Studies, Victoria University, Melbourne Silvers A (2000) Limited linkage, demand shifts and the transboundary transmission of regional growth. Reg Stud 34(3):239–251 Simmie J (2002) Trading places: competitive cities in the global economy. Eur Plan Stud 10(2):201–214 Simmie J (2004) Innovation and clustering in the globalised international economy. Urban Stud 41(5–6):1095–1112 Simmie J, Sennet J, Wood P, Hart D (2002) Innovation in Europe: a tale of networks, knowledge and trade in five cities. Reg Stud 36(1):47–64 Simon HA (1962) The architecture of complexity. Proc Am Philos Soc 106:467–482 Sit V, Liu W (2000) Restructuring and spatial change of China’s auto industry under institutional reform and globalization. Ann Assoc Am Geogr 90(4):653–673 Smith D, Timberlake M (2001) World city networks and hierarchies, 1977–1997. An empirical analysis of global air travel links. Am Behav Sci 44(10):1656–1678 Smith K (1994) Interactions in knowledge systems: foundations, policy implications and empirical methods. STEP rapport/report R-10 1994. Studies in Technology, Innovation and Economic Policy, Oslo Smith K (2000) What is the ‘knowledge economy’? knowledge-intensive industries and distributed knowledge bases. AEGIS Working Papers 2000–03, University of Western Sydney, Sydney Smith K (2002) Assessing the economic impacts of ICT. STEP Report Series R-01 2002. Studies in Technology, Innovation and Economic Policy, Oslo Sohn J (2002) Do birds of a feather flock together? economic linkage and geographic proximity. REAL Technical Papers 02-T-4, Regional Economics Applications Laboratory, Urbana-Champaign Sonis M, Hewings GJD (2001) Feedbacks in input-output systems: impacts, loops and hierarchies. In: Lahr ML, Dietzenbacher E (eds) Input-output analysis: frontiers and extensions. Palgrave, Hampshire, pp 71–99 Sonis M, Guilhoto J, Hewings GJD, Martins E (1995) Linkages, key sectors and structural change: some new perspectives. Dev Econ 33(3):233–270 Srholec M (2007) High-tech exports from developing countries: a symptom of technology spurts or statistical illusion? Rev World Econ 143(2):227–255 Staber U, Morrison C (1999) The empirical foundations of industrial district theory. Presented to the ISRN Workshop on Globalization and Regional Innovation Systems, 17–19 May 1999, Toronto STAR 21 (2002) Strategic aerospace review for the 21st century: Creating a coherent market and policy framework for a vital European Industry. European Commission Enterprise DirectorateGeneral: Aerospace, Defence, Rail and Maritime Industries Unit, Brussels Statistics Bureau of Japan (2007) Establishment and enterprise Census 2006http://www.stat.go.jp/ english/data/jigyou/2006/sokuhou/index.htm accessed 9 November 2007 Steinle C, Schiele H (2002) When do industries cluster? a proposal on how to assess an industry’s propensity to concentrate in a single region or nation. Res Policy 31(6):849–858 Storper M, Venables AJ (2004) Buzz: Face-to-face contact and the urban economy. J Econ Geogr 4(4):351–370 Storper M (1995) The resurgence of regional economies, ten years later. The region as a nexus of untraded interdependencies. Eur Urb Reg Stud 2(3):191–221 Storper M, Chen Y, De Paolis F (2002) Trade and the location of industries in the OECD and European union. J Econ Geogr 2(1):73–107 Sturgeon T (2003) What really goes on in Silicon Valley? spatial clustering and dispersal in modular production networks. J Econ Geogr 3:199–225

222

References

Sturgeon T, Florida R (1999) The world that changed the machine: globalization and jobs in the automotive industry. International Motor Vehicle Program: Final Report to the Alfred P. Sloan Foundation, Massachusetts Institute of Technology, Massachusetts Takeishi A, Fujimoto T (2003) Modularization in the car industry: inter-linked multiple hierarchies of product, production, and supplier systems. In: Prencipe A, Davies A, Hobday M (eds) The business of systems integration. Oxford University Press, Oxford, pp 254–278 Tan K, Khor H (2006) China’s changing economic structure and implications for regional patterns of trade, production and integration. China World Econ 14(6):1–19 Taylor PJ, Lang RE (2004) The shock of the new: 100 concepts describing recent urban change. Environ Plan A 36(6):951–958 Tellier L-N. (1997) A challenge for regional science: revealing and explaining the global spatial logic of economic development. Pap Reg Sci 76(4):371–384 Temple J (1999) The new growth evidence. J Econ Lit 37(1):112–157 Tether B, Metcalfe S (2003) Services and ‘systems of innovation’. CRIC Discussion Paper No 58, Centre for Research on Innovation and Competition, Manchester Tether B, Hipp C, Miles I (1999) Standardisation and specialisation in services: evidence from Germany. CRIC Discussion Paper No 30, Centre for Research on Innovation and Competition, Manchester Thompson JD (1967) Organizations in action: social science bases of administrative theory. McGraw Hill, New York Thompson K, Mitchell H, Knox S (1998) Organisational buying behavoiour in changing times. Eur Manage J 16(6):698–705 Tomlinson M (1997) The contribution of services to manufacturing: beyond the deindustrialisation debate. CRIC Discussion Paper No 5, Centre for Research on Innovation and Competition, Manchester Toner P (1999) Main currents in cumulative causation: the dynamics of growth and development. MacMillian, Basingstoke Torre A, Gilly JP (2000) On the analytical dimension of proximity dynamics. Reg Stud 34(2):169–180 Trend Business Research (2001) Business clusters in the UK – a first assessment. Department of Trade and Industry, London Turpin T (2000) Science and technology policies in asia-pacific economies: implications for regional knowledge systems and social transformation. CAPSTRANS - CEDA Policy Papers Series No 3, Committee for the Economic Development of Australia, Wollongong Tyson L (1992) Who’s bashing whom? trade conflict in high technology industries. Institute for International Economics, Washington UNDP (2001) Human development report 2001, making new technologies work for human development. Oxford University Press, Oxford: van den Berg L, Braun E, van Winden W (2001) Growth clusters in european cities: an integral approach. Urb Stud 38(1):185–205 van der Lann L, (1998) Changing urban systems: an empirical analysis at two spatial levels. Reg Stud 32(3):235–247 Van Pottelsberge De La Potterie B (1998) The efficiency of science and technology policies inside the triad. Dissertation, Universite De Bruxelles http://www.ulb.ac.be/cours/solvay/ vanpottelsberghe/index.html Venables AJ (1998) Trade and location: an assessment. Oxford Rev Econ Policy 14(2):1–6 Verbeek H (1999) Innovative clusters; identification of value adding production chains and their networks of innovation, an international study. Dissertation, http://www.oecd.org/ dataoecd/34/60/2098804.pdf Verspagen B, Schoenmakers W (2000) The spatial dimension of knowledge spillovers in Europe: evidence from firm patenting data. Presented at the AEA Conference on Intellectual Property Econometrics 19–20 April 2000, Alicante, Spain

References

223

Verspagen B (2002) Structural change and technology. A long view. Presented at the DRUID Summer Conference on Industrial Dynamics of the New and Old Economy - Who is Embracing Whom? Copenhagen/Elsinore 6–8 June 2002 Verspagen B, Schoenmakers W (2002) The spatial dimension of patenting by multinational firms in Europe. ECIS Working Paper No 02–11: Eindhoven Centre for Innovation Studies. The Netherlands Verspagen B, Wakelin K (1997) Trade and technology from a schumpeterian perspective. Int Rev Appl Econ 11(2):181–194 Verspagen B, Werker C (2003) The invisible college of the economics of innovation and technological change. MERIT-Infonomics Research Memorandum series 2003–08, Maastricht University, Maastricht Vinson T (2004) Community adversity and resilience: the distribution of social disadvantage in Victoria and New South Wales and the Mediating Role of Social Cohesion. The Ignatius Centre, Melbourne von Hippel E (1988) The sources of innovation. Oxford University Press, New York Voyer R (1999) Thirty years of Canadian Science Policy: From 1.5 to 1.5. Science Publ Policy 26(4):277–282 Wakelin K (1997) Trade and innovation: theory and evidence. Edward Elgar, Cheltenham Wakelin K (1998) The role of innovation in bilateral OECD Trade performance. Appl Econ 30:1335–1346 Wang Q, von Tunzelmann N (2000) Complexity and the functions of the firm: breadth and depth. Res Policy 29(7–8):805–818 Weintraub ER (1999) How should we write the history of twentieth-century economics. Oxford Rev Econ Policy 15(4):139–152 Wilkinson I (2001) A history of network and channels thinking in marketing in the 20th century. Austral J Market 9(2):23–53 Williamson J (2002) Winners and losers over two centuries of globalization. NBER Working Paper No 9161 http://www.nber.org/papers/w9161 Wixted B (2000) Supranational clusters: international trade, industrial interdependency and knowledge flows. Paper presented at the Thirteenth International Conference on Input-Output Techniques 21–25 August 2000, Macerata: Italy Wixted B, Cooper R (2002) Extraterritorial cluster development in Europe: searching for evidence of supranational industrial activity. Presented to the 14th International Conference on Input-output Techniques, 10–15 October 2002, Montreal Wixted B, Cooper R (2004) OECD inter-country cluster networks 1970–1990. Presented to the Regional Science Association International – World Congress 2004, 14–17 April 2004, Port Elizabeth: South Africa Wixted B (2005) Systems of innovation beyond borders: linked clustering and the role, scale and spatial structure of extra-territorial interdependencies. Dissertation, http://handle.uws.edu. au:8081/1959.7/639 Wixted B (2006) Cluster complexes: a framework for understanding the internationalisation of innovation systems. CPROST Reports 2006–04, Centre for Policy Research on Science and Technology, Vancouver Wixted B, Yamano N, Webb C (2006) Input-output analysis in an increasingly globalised world: applications of OECD’s Harmonised International Tables. OECD STI Working Paper 2006/7, OECD DSTI/DOC(2006)7, Paris Wixted B, Cooper R (2007) The evolution of OECD ICT inter-cluster networks 1970–2000: an input-output study of changes in the interdependencies between nine OECD economies. In: Cooper R, Donaghy K, Hewings GJD (eds) Advances in spatial science series: globalization and regional economic modeling. Springer, Berlin Wolf H (2000) International home bias in trade. Rev Econ Statist 82(4):555–563 Wolfe D, Lucas M (eds) (2004) Clusters in a cold climate: innovation dynamics in a diverse economy. Queen’s School of Policy Studies and McGill-Queen’s University Press, Kingston

224

References

Wolfe DA (2003) Clusters old and new: the transition to a knowledge economy in Canada’s Regions. Queen’s School of Policy Studies and McGill-Queen’s University Press, Kingston Wolfe DA, Gertler MS (2004) Clusters from the inside and out: local dynamics and global linkages economy. Urb Stud 41(5–6):1071–1093 Wolff E (2000) How persistent is industry specialization over time in industrialized countries. Int J Technol Manage 19(1–2):194–205 World Bank (1999) World Development Report 1999/2000 – Entering the 21st century: the changing development landscape. Oxford University Press, Washington World Commission on Environment and Development (1987) Our common future. Oxford University Press, Oxford World Trade Organization (2003) International trade statistics 2003. WTO, Geneva World Trade Organization (2006) International trade statistics 2003. WTO, Geneva Wyckoff A (1993) The extension of networks of production across borders. OECD STI Review No. 13:61–87 Yamano N, Ahmad N (2006) The OECD input-output database: 2006 edition. STI Working Paper 2006/8, Paris: OECD DSTI/DOC(2006)8 pp1–60 Yeats AJ (1998) Just how big is global production sharing? World Bank Working Papers No 1871, World Bank, Washington Yetton P, Craig J, Davis J, Hilmer F (1992) Are diamonds a country’s best friend? a critique of porter’s theory of national competition as applied to Canada, New Zealand and Australia. Aust J Manage 17(1):89–119 Zysman J, Doherty E, Schwartz A (1996) Tales from the ‘global’ economy: cross nation production networks and the re-organization of the European Economy. Berkeley Roundtable on the International Economy Working Paper No 83, BRIE, Berkeley, http://brie. berkeley.edu/~briewww/

Index

A absorptive capacity, 25 agglomeration, 4 Airbus, 6, 84, 85, 86, 87, 148, 156, 157, 158, 159, 160, 161, 162 Apple, 169 Asia-Pacific Americas(APAs), 156, 178, 199 Australia, 6, 9, 15, 20, 21, 25, 65, 75, 76, 83, 84, 85, 86, 94, 149, 151, 155, 156, 157, 160, 161, 170, 173, 175, 178, 181, 185, 186 Austria, 65, 158, 159, 178, 181

E East Asia, 4 embeddedness, 5 Embraer, 148 exo-nets, 149, 156, 178, 197

B Boeing, 6, 84, 85, 86, 87, 148, 157, 158, 162 Bombardier, 148

G Germany, 24, 27, 31, 65, 75, 93, 96, 103, 104, 111, 149, 151, 155, 156, 157, 158, 159, 160, 161, 173, 175, 178, 181, 186, 190, 196, 199, 200 globalisation, 19, 198 global production networks, 182

C Canada, 3, 6, 13, 16, 65, 70, 75, 81, 96, 97, 111, 126, 129, 149, 151, 155, 156, 158, 169, 171, 173, 175, 178, 181, 185, 186, 197 China, 1, 12, 32, 66, 105, 151, 158, 170, 175, 178, 181, 183, 185, 186, 187, 189, 190 comparative, 16

D Dell, 182 Denmark, 65, 74, 149, 151, 155, 156, 158, 173, 175, 178, 181

F flying geese model, 183 fragmentation, 2, 4, 5, 9, 12, 31, 59, 72, 73, 78, 95, 113, 178, 182, 183, 190, 196, 197, 202, 204

I IBM, 182 Iceland, 5, 65 ICT, 166 interdependencies, 6, 7, 8, 9, 10, 11, 14, 18, 23, 32, 59, 61, 66, 67, 76, 77, 79, 80, 81, 82, 83, 85, 87, 89, 90, 92, 95, 96, 107, 161, 165, 178, 183, 187, 192, 193, 194, 195, 196, 197, 198, 199, 201 international value systems, 162 Ireland, 65, 111, 158, 167, 170, 181, 185, 186, 187, 188, 189, 190, 196 Israel, 3

225

226

Index

J Japan, 13, 15, 18, 27, 62, 65, 66, 74, 75, 81, 93, 96, 97, 129, 149, 151, 155, 156, 158, 159, 160, 162, 167, 168, 169, 170, 173, 175, 178, 181, 183, 185, 186, 190, 196, 199, 200

R Rolls Royce, 147 R&D, 1, 9, 13, 17, 20, 22, 23, 25, 26, 62, 66, 79, 80, 81, 84, 86, 87, 92, 105, 113, 114, 147, 162, 195 intensity, 114

M macro-modules, 161 Malaysia, 1 Mexico, 1 multi-spatial systems of innovation, 107, 161

S Scandinavia, 3 Silicon Valley, 168, 195 Singapore, 1, 66, 169, 170, 186 South Korea, 1, 105, 185, 186, 200 Sweden, 91, 158 Switzerland, 65

N New Zealand, 65, 75 Non-ferrous metals, 111

T Taiwan, 1, 3, 66, 80, 178, 181, 185, 186, 190, 195, 200

P Petroleum, 111, 112, 114, 125, 165, 195

U United Kingdom, 3, 6, 178 United States of America, 5, 28, 155